Search NASASearch

SEARCH · Search NASA

Results for “Schwarz methods”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 37 records · Page 2

Two-level overlapping additive Schwarz preconditioner for training scientific machine learning applications

In this work we introduce a novel two-level overlapping additive Schwarz preconditioner for accelerating the training of scientific machine learning applications. The design of the proposed preconditioner is motivated by the nonlinear two-level overlapping additive Schwarz preconditioner. The neural network parameters are decomposed into groups (subdomains) with overlapping regions. In addition, the network’s feed-forward structure is indirectly imposed through a novel subdomain-wise synchronization strategy and a coarse-level training step. Through a series of numerical experiments, which consider physicsinformed neural networks and operator learning approaches, we demonstrate that the proposed two-level preconditioner significantly speeds up the convergence of the standard (LBFGS) optimizer while also yielding more accurate machine learning models. Moreover, the devised preconditioner is designed to take advantage of model-parallel computations, which can further reduce the training time.

97 MATHEMATICS AND COMPUTING

Experiments with elliptic problem solvers on the Cedar multicluster

The implementation and performance of algorithms for the solution of elliptic problems on the Cedar multiprocessor is examined. The algorithms considered are conjugate gradient schemes with or without preconditioning on a rectangular domain and domain decomposition techniques with overlapping (Schwarz Alternating Method) and nonoverlapping subdomains on a T-shaped domain. Some preliminary results from a current research effort are presented.

Frank, G.

Alternating method applied to edge and surface crack problems

The Schwarz-Neumann alternating method is employed to obtain stress intensity solutions to two crack problems of practical importance: a semi-infinite elastic plate containing an edge crack which is subjected to concentrated normal and tangential forces, and an elastic half space containing a semicircular surface crack which is subjected to uniform opening pressure. The solution to the semicircular surface crack is seen to be a significant improvement over existing approximate solutions. Application of the alternating method to other crack problems of current interest is briefly discussed.

Hartranft, R. J.

A multi-domain method for subsonic viscous flows

We have developed a Schwarz type domain decomposition method for a pressure base, two- and three-dimensional Navier-Stokes solver. This technique allows one to partition a flow path, which can be characterized by complex geometry and/or complicated flow physics, into smaller sub-domains according to the local geometric simplicity or estimated flow scales. We can, then, sweep the sub-domains in some order and solve the Navier-Stokes equations using as boundary conditions, along the domain interfaces, the Dirichlet conditions which are taken from the most recent update of the solution in the adjacent neighboring domains. With this technique, one can minimize the adverse effects caused by grid skewness and the stiffness problem caused by disparate flow scales. Here, we report the results of a few fundamental flow cases to demonstrate that a judicious use of the multi-domain method can offer a significant convergence acceleration over the traditional one-domain method. This method can be extended to exploit the architecture of a parallel computer to further improve the speed.

Chan, Daniel C.

Parallel Implicit Algorithms for CFD

The main goal of this project was efficient distributed parallel and workstation cluster implementations of Newton-Krylov-Schwarz (NKS) solvers for implicit Computational Fluid Dynamics (CFD.) "Newton" refers to a quadratically convergent nonlinear iteration using gradient information based on the true residual, "Krylov" to an inner linear iteration that accesses the Jacobian matrix only through highly parallelizable sparse matrix-vector products, and "Schwarz" to a domain decomposition form of preconditioning the inner Krylov iterations with primarily neighbor-only exchange of data between the processors. Prior experience has established that Newton-Krylov methods are competitive solvers in the CFD context and that Krylov-Schwarz methods port well to distributed memory computers. The combination of the techniques into Newton-Krylov-Schwarz was implemented on 2D and 3D unstructured Euler codes on the parallel testbeds that used to be at LaRC and on several other parallel computers operated by other agencies or made available by the vendors. Early implementations were made directly in Massively Parallel Integration (MPI) with parallel solvers we adapted from legacy NASA codes and enhanced for full NKS functionality. Later implementations were made in the framework of the PETSC library from Argonne National Laboratory, which now includes pseudo-transient continuation Newton-Krylov-Schwarz solver capability (as a result of demands we made upon PETSC during our early porting experiences). A secondary project pursued with funding from this contract was parallel implicit solvers in acoustics, specifically in the Helmholtz formulation. A 2D acoustic inverse problem has been solved in parallel within the PETSC framework.

Keyes, David E.

The Multigrid-Mask Numerical Method for Solution of Incompressible Navier-Stokes Equations

A multigrid-mask method for solution of incompressible Navier-Stokes equations in primitive variable form has been developed. The main objective is to apply this method in conjunction with the pseudospectral element method solving flow past multiple objects. There are two key steps involved in calculating flow past multiple objects. The first step utilizes only Cartesian grid points. This homogeneous or mask method step permits flow into the interior rectangular elements contained in objects, but with the restriction that the velocity for those Cartesian elements within and on the surface of an object should be small or zero. This step easily produces an approximate flow field on Cartesian grid points covering the entire flow field. The second or heterogeneous step corrects the approximate flow field to account for the actual shape of the objects by solving the flow field based on the local coordinates surrounding each object and adapted to it. The noise occurring in data communication between the global (low frequency) coordinates and the local (high frequency) coordinates is eliminated by the multigrid method when the Schwarz Alternating Procedure (SAP) is implemented. Two dimensional flow past circular and elliptic cylinders will be presented to demonstrate the versatility of the proposed method. An interesting phenomenon is found that when the second elliptic cylinder is placed in the wake of the first elliptic cylinder a traction force results in a negative drag coefficient.

Ku, Hwar-Ching

Residual interference and wind tunnel wall adaption

Measured flow variables near the test section boundaries, used to guide adjustments of the walls in adaptive wind tunnels, can also be used to quantify the residual interference. Because of a finite number of wall control devices (jacks, plenum compartments), the finite test section length, and the approximation character of adaptation algorithms, the unconfined flow conditions are not expected to be precisely attained even in the fully adapted stage. The procedures for the evaluation of residual wall interference are essentially the same as those used for assessing the correction in conventional, non-adaptive wind tunnels. Depending upon the number of flow variables utilized, one can speak of one- or two-variable methods; in two dimensions also of Schwarz- or Cauchy-type methods. The one-variable methods use the measured static pressure and normal velocity at the test section boundary, but do not require any model representation. This is clearly of an advantage for adaptive wall test section, which are often relatively small with respect to the test model, and for the variety of complex flows commonly encountered in wind tunnel testing. For test sections with flexible walls the normal component of velocity is given by the shape of the wall, adjusted for the displacement effect of its boundary layer. For ventilated test section walls it has to be measured by the Calspan pipes, laser Doppler velocimetry, or other appropriate techniques. The interface discontinuity method, also described, is a genuine residual interference assessment technique. It is specific to adaptive wall wind tunnels, where the computation results for the fictitious flow in the exterior of the test section are provided.

Mokry, Miroslav

Schwarz-Based Algorithms for Compressible Flows

We investigate in this paper the application of Schwarz-based algorithms to compressible flows. First we study the combination of these methods with defect-correction procedures. We then study the effect on the Schwarz-based methods of replacing the explicit treatment of the boundary conditions by an implicit one. In the last part of this paper we study the combination of these methods with Newton-Krylov matrix-free methods. Numerical experiments that show the performance of our approaches are then presented.

Tidriri, M. D.

Globalized Newton-Krylov-Schwarz Algorithms and Software for Parallel Implicit CFD

Implicit solution methods are important in applications modeled by PDEs with disparate temporal and spatial scales. Because such applications require high resolution with reasonable turnaround, "routine" parallelization is essential. The pseudo-transient matrix-free Newton-Krylov-Schwarz (Psi-NKS) algorithmic framework is presented as an answer. We show that, for the classical problem of three-dimensional transonic Euler flow about an M6 wing, Psi-NKS can simultaneously deliver: globalized, asymptotically rapid convergence through adaptive pseudo- transient continuation and Newton's method-, reasonable parallelizability for an implicit method through deferred synchronization and favorable communication-to-computation scaling in the Krylov linear solver; and high per- processor performance through attention to distributed memory and cache locality, especially through the Schwarz preconditioner. Two discouraging features of Psi-NKS methods are their sensitivity to the coding of the underlying PDE discretization and the large number of parameters that must be selected to govern convergence. We therefore distill several recommendations from our experience and from our reading of the literature on various algorithmic components of Psi-NKS, and we describe a freely available, MPI-based portable parallel software implementation of the solver employed here.

Gropp, W. D.

Local multiplicative Schwarz algorithms for convection-diffusion equations

We develop a new class of overlapping Schwarz type algorithms for solving scalar convection-diffusion equations discretized by finite element or finite difference methods. The preconditioners consist of two components, namely, the usual two-level additive Schwarz preconditioner and the sum of some quadratic terms constructed by using products of ordered neighboring subdomain preconditioners. The ordering of the subdomain preconditioners is determined by considering the direction of the flow. We prove that the algorithms are optimal in the sense that the convergence rates are independent of the mesh size, as well as the number of subdomains. We show by numerical examples that the new algorithms are less sensitive to the direction of the flow than either the classical multiplicative Schwarz algorithms, and converge faster than the additive Schwarz algorithms. Thus, the new algorithms are more suitable for fluid flow applications than the classical additive or multiplicative Schwarz algorithms.

Cai, Xiao-Chuan

Classical free-streamline flow over a polygonal obstacle

In classical Kirchhoff flow, an ideal incompressible fluid flows past an obstacle and around a motionless wake bounded by free streamlines. Since 1869 it has been known that in principle, the two-dimensional Kirchhoff flow over a polygonal obstacle can be determined by constructing a conformal map onto a polygon in the log-hodograph plane. In practice, however, this idea has rarely been put to use except for very simple obstacles, because the conformal mapping problem has been too difficult. This paper presents a practical method for computing flows over arbitrary polygonal obstacles to high accuracy in a few seconds of computer time. We achieve this high speed and flexibility by working with a modified Schwarz-Christoffel integral that maps onto the flow region directly rather than onto the log-hodograph polygon. This integral and its associated parameter problem are treated numerically by methods developed earlier by Trefethen for standard Schwarz-Christoffel maps.

Elcrat, A. R.

Classical free-streamline flow over a polygonal obstacle

In classical Kirchhoff flow, an ideal incompressible fluid flows past an obstacle and around a motionless wake bounded by free streamlines. Since 1869 it has been known that in principle, the two-dimensional Kirchhoff flow over a polygonal obstacle can be determined by constructing a conformal map onto a polygon in the log-hodograph plane. In practice, however, this idea has rarely been put to use except for very simple obstacles, because the conformal mapping problem has been too difficult. This paper presents a practical method for computing flows over arbitrary polygonal obstacles to high accuracy in a few seconds of computer time. We achieve this high speed and flexibility by working with a modified Schwarz-Christoffel integral that maps onto the flow region directly rather than onto the log-hodograph polygon. This integral and its associated parameter problem are treated numerically by methods developed earlier by Trefethen for standard Schwarz-Christoffel maps.

Elcrat, A. R.

Two-dimensional wind-tunnel interference from measurements on two contours

This paper describes how wall-induced velocities near a model in a two-dimensional wind tunnel can be estimated from upwash distributions measured along two contours surrounding a model. The method is applicable to flows that can be represented by linear theory. It was derived by applying the Schwarz Integral Formula separately to the two contours and by exploiting the free-air relationship between upwashes along the contours. Advantages of the method are that only one flow quantity need by measured and no representation of the model is required. A weakness of the method is that it assumes streamwise interference velocity vanishes far upstream of the model. This method was applied to a simple theoretical model of flow in a solid-wall wind tunnel. The theoretical interference velocities and the velocities computed using the method were in excellent agreement. The method was then used to analyze experimental data acquired during adaptive-wall experiments at Ames Research Center. This analysis confirmed that the wall adjustments reduced wall-induced velocities near the model.

Schairer, E. T.

Calculations of Diffuser Flows with an Anisotropic K-Epsilon Model

A newly developed anisotropic K-epsilon model is applied to calculate three axisymmetric diffuser flows with or without separation. The new model uses a quadratic stress-strain relation and satisfies the realizability conditions, i.e., it ensures both the positivity of the turbulent normal stresses and the Schwarz' inequality between any fluctuating velocities. Calculations are carried out with a finite-element method. A second-order accurate, bounded convection scheme and sufficiently fine grids are used to ensure numerical credibility of the solutions. The standard K-epsilon model is also used in order to highlight the performance of the new model. Comparison with the experimental data shows that the anisotropic K-epsilon model performs consistently better than does the standard K-epsilon model in all of the three test cases.

Zhu, J.

NASA Bioreactors Advance Disease Treatments

The International Space Station (ISS) is falling. This is no threat to the astronauts onboard, however, because falling is part of the ISS staying in orbit. The absence of gravity beyond the Earth s atmosphere is actually an illusion; at the ISS s orbital altitude of approximately 250 miles above the surface, the planet s gravitational pull is only 12-percent weaker than on the ground. Gravity is constantly pulling the ISS back to Earth, but the space station is also constantly traveling at nearly 18,000 miles per hour. This means that, even though the ISS is falling toward Earth, it is moving sideways fast enough to continually miss impacting the planet. The balance between the force of gravity and the ISS s motion creates a stable orbit, and the fact that the ISS and everything in it including the astronauts are falling at an equal rate creates the condition of weightlessness called microgravity. The constant falling of objects in orbit is not only an important principle in space, but it is also a key element of a revolutionary NASA technology here on Earth that may soon help cure medical ailments from heart disease to diabetes. In the mid-1980s, NASA researchers at Johnson Space Center were investigating the effects of long-term microgravity on human tissues. At the time, the Agency s shuttle fleet was grounded following the 1986 Space Shuttle Challenger disaster, and researchers had no access to the microgravity conditions of space. To provide a method for recreating such conditions on Earth, Johnson s David Wolf, Tinh Trinh, and Ray Schwarz developed that same year a horizontal, rotating device called a rotating wall bioreactor that allowed the growth of human cells in simulated weightlessness. Previously, cell cultures on Earth could only be grown two-dimensionally in Petri dishes, because gravity would cause the multiplying cells to sink within their growth medium. These cells do not look or function like real human cells, which grow three-dimensionally in the body. Experiments conducted by Johnson scientist Dr. Thomas Goodwin proved that the NASA bioreactor could successfully cultivate cells using simulated microgravity, resulting in three-dimensional tissues that more closely approximate those in the body. Further experiments conducted on space shuttle missions and by Wolf as an astronaut on the Mir space station demonstrated that the bioreactor s effects were even further expanded in space, resulting in remarkable levels of tissue formation. While the bioreactor may one day culture red blood cells for injured astronauts or single-celled organisms like algae as food or oxygen producers for a Mars colony, the technology s cell growth capability offers significant opportunities for terrestrial medical research right now. A small Texas company is taking advantage of the NASA technology to advance promising treatment applications for diseases both common and obscure.

Source record