Augmented Lagrangian Scheme Research Articles

A Surface-to-Surface Finite Element Algorithm for Large Deformation Frictional Contact in febio.

This study formulates a finite element algorithm for frictional contact of solid materials, accommodating finite deformation and sliding. The algorithm uses a penalty method regularized with an augmented Lagrangian scheme to enforce contact constraints in a nonmortar surface-to-surface approach. Use of a novel kinematical approach to contact detection and enforcement of frictional constraints allows solution of complex problems previously requiring mortar methods or contact smoothing algorithms. Patch tests are satisfied to a high degree of accuracy with a single-pass penalty method, ensuring formulation errors do not affect the solution. The accuracy of the implementation is verified with Hertzian contact, and illustrations demonstrating the ability to handle large deformations and sliding are presented and validated against prior literature. A biomechanically relevant example addressing finger friction during grasping demonstrates the utility of the proposed algorithm. The algorithm is implemented in the open source software febio, and the source code is made available to the general public.

Nonconvex and Nonsmooth Optimization with Generalized Orthogonality Constraints: An Approximate Augmented Lagrangian Method

Nonconvex and nonsmooth optimization problems with linear equation and generalized orthogonality constraints have wide applications. These problems are difficult to solve due to nonsmooth objective function and nonconvex constraints. In this paper, by introducing an extended proximal alternating linearized minimization (EPALM) method, we propose a framework based on the augmented Lagrangian scheme (EPALMAL). We also show that the EPALMAL method has global convergence in the sense that every bounded sequence generated by the EPALMAL method has at least one convergent subsequence that converges to the Karush---Kuhn---Tucker point of the original problem. Experiments on a variety of applications, including compressed modes and multivariate data analysis, have demonstrated that the proposed method is noticeably efficient and achieves comparable performance with existing methods.

Unimodular Sequence Design Based on Alternating Direction Method of Multipliers

The topic of probing waveform design has received considerable attention due to its numerous applications in active sensing. Apart from having the desirable property of constant magnitude, it is also anticipated that the designed sequence possesses low sidelobe autocorrelation and/or specified spectral shape. In this paper, the alternating direction method of multipliers (ADMM), which is a powerful variant of the augmented Lagrangian scheme for dealing with separable objective functions, is applied for synthesizing the probing sequences. To achieve impulse-like autocorrelation, we formulate the design problem as minimizing a nonlinear least-squares cost function in the frequency domain subject to the constraint that all sequence elements are of unit modulus. Via introducing auxiliary variables, we are able to separate the objective into linear and quadratic functions where the unimodular constraint is only imposed on the former, which results in an ADMM-style iterative procedure. In particular, fast implementation for the most computationally demanding step is investigated and local convergence of the ADMM method is proved. To deal with the spectral shape requirement, we borrow the concept in frequency-selective filter design where passband and stopband magnitudes are bounded to formulate the corresponding optimization problem. In this ADMM algorithm development, unit-step functions are utilized to transform the multivariable optimization into a quadratic polynomial problem with a single variable. The effectiveness of the proposed approach is demonstrated via computer simulations.

Two–dimensional viscoplastic dambreaks

We report the results of computations for two–dimensional dambreaks of viscoplastic fluid, focusing on the phenomenology of the collapse, the mode of initial failure, and the final shape of the slump. The volume-of-fluid method is used to evolve the surface of the viscoplastic fluid, and its rheology is captured by either regularizing the viscosity or using an augmented-Lagrangian scheme. We outline a modification to the volume-of-fluid scheme that eliminates resolution problems associated with the no-slip condition applied on the underlying surface. We establish that the regularized and augmented-Lagrangian methods yield comparable results, except for the stress field at the initiation or termination of motion. The numerical results are compared with asymptotic theories valid for relatively shallow or vertically slender flow, with a series of previously reported experiments, and with predictions based on plasticity theory.

Sequential equality-constrained optimization for nonlinear programming

A novel idea is proposed for solving optimization problems with equality constraints and bounds on the variables. In the spirit of sequential quadratic programming and sequential linearly-constrained programming, the new proposed approach approximately solves, at each iteration, an equality-constrained optimization problem. The bound constraints are handled in outer iterations by means of an augmented Lagrangian scheme. Global convergence of the method follows from well-established nonlinear programming theories. Numerical experiments are presented.

Fast iterative algorithm for the reconstruction of multishot non-cartesian diffusion data.

To accelerate the motion-compensated iterative reconstruction of multishot non-Cartesian diffusion data. The motion-compensated recovery of multishot non-Cartesian diffusion data is often performed using a modified iterative sensitivity-encoded algorithm. Specifically, the encoding matrix is replaced with a combination of nonuniform Fourier transforms and composite sensitivity functions, which account for the motion-induced phase errors. The main challenge with this scheme is the significantly increased computational complexity, which is directly related to the total number of composite sensitivity functions (number of shots × number of coils). The dimensionality of the composite sensitivity functions and hence the number of Fourier transforms within each iteration is reduced using a principal component analysis-based scheme. Using a Toeplitz-based conjugate gradient approach in combination with an augmented Lagrangian optimization scheme, a fast algorithm is proposed for the sparse recovery of diffusion data. The proposed simplifications considerably reduce the computation time, especially in the recovery of diffusion data from under-sampled reconstructions using sparse optimization. By choosing appropriate number of basis functions to approximate the composite sensitivities, faster reconstruction (close to 9 times) with effective motion compensation is achieved. The proposed enhancements can offer fast motion-compensated reconstruction of multishot diffusion data for arbitrary k-space trajectories.

Interior Epigraph Directions method for nonsmooth and nonconvex optimization via generalized augmented Lagrangian duality

We propose and study a new method, called the Interior Epigraph Directions (IED) method, for solving constrained nonsmooth and nonconvex optimization. The IED method considers the dual problem induced by a generalized augmented Lagrangian duality scheme, and obtains the primal solution by generating a sequence of iterates in the interior of the dual epigraph. First, a deflected subgradient (DSG) direction is used to generate a linear approximation to the dual problem. Second, this linear approximation is solved using a Newton-like step. This Newton-like step is inspired by the Nonsmooth Feasible Directions Algorithm (NFDA), recently proposed by Freire and co-workers for solving unconstrained, nonsmooth convex problems. We have modified the NFDA so that it takes advantage of the special structure of the epigraph of the dual function. We prove that all the accumulation points of the primal sequence generated by the IED method are solutions of the original problem. We carry out numerical experiments by using test problems from the literature. In particular, we study several instances of the Kissing Number Problem, previously solved by various approaches such as an augmented penalty method, the DSG method, as well as several popular differentiable solvers. Our experiments show that the quality of the solutions obtained by the IED method is comparable with (and sometimes favourable over) those obtained by the differentiable solvers.

A fixed-point augmented Lagrangian method for total variation minimization problems

In this paper, we propose a fixed-point augmented Lagrangian method (FPALM) for general convex problems arising in image processing. We can easily obtain the alternating minimization algorithm (AMA) referred to [1] from the proposed FPALM. The proof for the convergence of the FPALM is provided under some mild assumptions. We present two kinds of first-order augmented Lagrangian schemes and show their connections to first-order primal–dual algorithms [2]. Furthermore, we apply an acceleration rule to both the FPALM and AMA to achieve better convergence rates. Numerical examples on different image denosing models including the ROF model, the vectorial TVmodel, high order models and the TV-L1 model are provided to demonstrate the efficiency of the proposed algorithms.

Smoothly Clipped Absolute Deviation (SCAD) regularization for compressed sensing MRI Using an augmented Lagrangian scheme

PurposeCompressed sensing (CS) provides a promising framework for MR image reconstruction from highly undersampled data, thus reducing data acquisition time. In this context, sparsity-promoting regularization techniques exploit the prior knowledge that MR images are sparse or compressible in a given transform domain. In this work, a new regularization technique was introduced by iterative linearization of the non-convex smoothly clipped absolute deviation (SCAD) norm with the aim of reducing the sampling rate even lower than it is required by the conventional l1 norm while approaching an l0 norm. Materials and MethodsThe CS-MR image reconstruction was formulated as an equality-constrained optimization problem using a variable splitting technique and solved using an augmented Lagrangian (AL) method developed to accelerate the optimization of constrained problems. The performance of the resulting SCAD-based algorithm was evaluated for discrete gradients and wavelet sparsifying transforms and compared with its l1-based counterpart using phantom and clinical studies. The k-spaces of the datasets were retrospectively undersampled using different sampling trajectories. In the AL framework, the CS-MRI problem was decomposed into two simpler sub-problems, wherein the linearization of the SCAD norm resulted in an adaptively weighted soft thresholding rule with a sparsity enhancing effect. ResultsIt was demonstrated that the proposed regularization technique adaptively assigns lower weights on the thresholding of gradient fields and wavelet coefficients, and as such, is more efficient in reducing aliasing artifacts arising from k-space undersampling, when compared to its l1-based counterpart. ConclusionThe SCAD regularization improves the performance of l1-based regularization technique, especially at reduced sampling rates, and thus might be a good candidate for some applications in CS-MRI.

Non-linear extension of FFT-based methods accelerated by conjugate gradients to evaluate the mechanical behavior of composite materials

FFT-based methods are used to solve the problem of a heterogeneous unit-cell submitted to periodic boundary conditions, which is of a great interest in the context of numerical homogenization. Recently (in 2010), Brisard and Zeman proposed simultaneously to use Conjugate Gradient based solvers in order to improve the convergence properties (when compared to the basic scheme, proposed initially in 1994). The purpose of the paper is to extend this idea to the case of non-linear behaviors. The proposed method is based on a Newton–Raphson algorithm and can be applied to various kinds of behaviors (time dependant or independent, with or without internal variables) through a conventional integration procedure as used in finite element codes. It must be pointed out that this approach is fundamentally different from the traditional FFT-based approaches which rely on a fixed-point algorithm (e.g. basic scheme, Eyre and Milton accelerated scheme, Augmented Lagrangian scheme, etc.). The method is compared to the basic scheme on the basis of a simple application (a linear elastic spherical inclusion within a non-linear elastic matrix): a low sensitivity to the reference material and an improved efficiency, for a soft or a stiff inclusion, are observed. At first proposed for a prescribed macroscopic strain, the method is then extended to mixed loadings.

Augmented Lagrangian method for a mean curvature based image denoising model

High order derivative information has been widely used in developing variational models in image processing to accomplish more advanced tasks. However, it is a nontrivial issue to construct efficient numerical algorithms to deal with the minimization of these variational models due to the associated high order Euler-Lagrange equations. In this paper, we propose an efficient numerical method for a mean curvature based image denoising model using the augmented Lagrangian method. A special technique is introduced to handle the mean curvature model for the augmented Lagrangian scheme. We detail the procedures of finding the related saddle-points of the functional. We present numerical experiments to illustrate the effectiveness and efficiency of the proposed numerical method, and show a few important features of the image denoising model such as keeping corners and image contrast. Moreover, a comparison with the gradient descent method further demonstrates the efficiency of the proposed augmented Lagrangian method.

Isothermal dynamics of channeled viscoplastic lava flows and new methods for estimating lava rheology

This study analyses the influence of a viscoplastic lava rheology on the dynamics of lava flows. Using a multigrid‐based augmented Lagrangian scheme, we find a numerical solution for the flow of a Bingham fluid in a rectangular channel. The numerical results show that an internal viscoplastic rheology significantly modifies the velocity distribution within a lava flow through the development of plug regions whose size is determined by the magnitude of the yield strength. The flow rate, maximum surface velocity and central plug dimensions are determined as functions of the channel geometry and fluid rheology, and comparisons between these and several limiting analytical solutions confirm the accuracy of the numerical method used. The results are also compared to incorrect models which have been proposed previously in the literature. Several algorithms that extend the results to different sets of measured initial parameters are outlined; these calculate: (1) the flow depth when the fluid rheology (viscosity and yield strength) and downstream flow rate are given, (2) the flow depth when the fluid rheology and maximum downstream surface velocity are given, (3) the flow rate and fluid rheology when the flow depth, maximum surface velocity and surface plug width are given, and (4) the flow depth and rheology when the flow rate, maximum surface velocity and surface plug width are given. The use of these algorithms is demonstrated by considering the dynamics of a typical lava flow on Mount Etna, using measured rheological parameters and field observations.

On the convergence of augmented Lagrangian methods for nonlinear semidefinite programming

In this paper, we present new convergence properties of the augmented Lagrangian method for nonlinear semidefinite programs (NSDP). Convergence to the approximately global solutions and optimal values of NSDP is first established for a basic augmented Lagrangian scheme under mild conditions, without requiring the boundedness condition of the multipliers. We then propose four modified augmented Lagrangian methods for NSDP based on different algorithmic strategies. We show that the same convergence of the proposed methods can be ensured under weaker conditions.

On primal convergence for augmented Lagrangian duality

We study some theoretical properties of augmented Lagrangian duality schemes for nonconvex and nonsmooth optimization. These schemes may not have primal convergence, in the sense that limit points of the primal sequence may not solve the primal problem. First, we prove that primal convergence holds if and only if the dual function is differentiable at the dual limit. Second, we relate the latter differentiability property with the existence of exact penalty parameters. Finally, we give a condition on the Lagrangian which ensures that every point supports an exact penalty representation.

An augmented penalty function method with penalty parameter updates for nonconvex optimization

Given an augmented Lagrangian scheme for a general optimization problem, we use an epsilon subgradient step for improving the dual function. This can be seen as an update for an augmented penalty method, which is more stable because it does not force the penalty parameter to tend to infinity. We establish for this update primal–dual convergence for our augmented penalty method. As illustration, we apply our method to the test-bed kissing number problem.

An augmented Lagrangian approach with a variable transformation in nonlinear programming

Tangent cone and (regular) normal cone of a closed set under an invertible variable transformation around a given point are investigated, which lead to the concepts of θ − 1 -tangent cone of a set and θ − 1 -subderivative of a function. When the notion of θ − 1 -subderivative is applied to perturbation functions, a class of augmented Lagrangians involving an invertible mapping of perturbation variables are obtained, in which dualizing parameterization and augmenting functions are not necessarily convex in perturbation variables. A necessary and sufficient condition for the exact penalty representation under the proposed augmented Lagrangian scheme is obtained. For an augmenting function with an Euclidean norm, a sufficient condition (resp., a sufficient and necessary condition) for an arbitrary vector (resp., 0) to support an exact penalty representation is given in terms of θ − 1 -subderivatives. An example of the variable transformation applied to constrained optimization problems is given, which yields several exact penalization results in the literature.

The Zero Duality Gap Property and Lower Semicontinuity of the Perturbation Function

We examine the validity of the zero duality gap properties for two important dual schemes: a generalized augmented Lagrangian dual scheme and a nonlinear Lagrange-type dual scheme. The necessary and sufficient conditions for the zero duality gap property to hold are established in terms of the lower semicontinuity of the perturbation functions.

Simultaneous structural analysis and design based on augmented Lagrangian duality

Solution procedures in structural optimization are commonly based on a nested approach where approximations of the analysis and design problems are solved alternately in an iterative scheme. In this paper, we study a simultaneous approach based on an integrated formulation of the analysis and design problems. An advantage of the simultaneous approach, when compared to the nested one, is that the dependence between the analysis and design variables is imposed explicitly. In the nested approach, this dependence is implicitly determined through the solution of the analysis problem. Earlier simultaneous approaches mostly utilize various penalty function reformulations. In this paper, we make use of two augmented Lagrangian schemes, which avoid the numerical ill-conditioning inherent in penalty reformulations. These schemes give rise to Lagrangian subproblems with somewhat different properties, and two efficient techniques are adapted for their solution. The first is a projected Newton method, and the second is a simplicial decomposition scheme. Computational results for bar-truss structures show that the proposed schemes are viable approaches for solving the integrated formulation, and that they are promising for future developments.