Search NASA⌕ Search

SEARCH · Search NASA

Results for “direct solver”

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 325 records · Page 18

A Fast Parallel Simulation Code for Interaction between Proto-Planetary Disk and Embedded Proto-Planets: Implementation for 3D Code

We develop a 3D simulation code for interaction between the proto-planetary disk and embedded proto-planets. The protoplanetary disk is treated as a three-dimensional (3D), self-gravitating gas whose motion is described by the locally isothermal Navier-Stokes equations in a spherical coordinate centered on the star. The differential equations for the disk are similar to those given in Kley et al. (2009) with a different gravitational potential that is defined in Nelson et al. (2000). The equations are solved by directional split Godunov method for the inviscid Euler equations plus operator-split method for the viscous source terms. We use a sub-cycling technique for the azimuthal sweep to alleviate the time step restriction. We also extend the FARGO scheme of Masset (2000) and modified in Li et al. (2001) to our 3D code to accelerate the transport in the azimuthal direction. Furthermore, we have implemented a reduced 2D (r, {theta}) and a fully 3D self-gravity solver on our uniform disk grid, which extends our 2D method (Li, Buoni, & Li 2008) to 3D. This solver uses a mode cut-off strategy and combines FFT in the azimuthal direction and direct summation in the radial and meridional direction. An initial axis-symmetric equilibrium disk is generated via iteration between the disk density profile and the 2D disk-self-gravity. We do not need any softening in the disk self-gravity calculation as we have used a shifted grid method (Li et al. 2008) to calculate the potential. The motion of the planet is limited on the mid-plane and the equations are the same as given in D'Angelo et al. (2005), which we adapted to the polar coordinates with a fourth-order Runge-Kutta solver. The disk gravitational force on the planet is assumed to evolve linearly with time between two hydrodynamics time steps. The Planetary potential acting on the disk is calculated accurately with a small softening given by a cubic-spline form (Kley et al. 2009). Since the torque is extremely sensitive to the position of the planet, we adopt the corotating frame that allows the planet moving only in radial direction if only one planet is present. This code has been extensively tested on a number of problems. For the earthmass planet with constant aspect ratio h = 0.05, the torque calculated using our code matches quite well with the the 3D linear theory results by Tanaka et al. (2002). The code is fully parallelized via message-passing interface (MPI) and has very high parallel efficiency. Several numerical examples for both fixed planet and moving planet are provided to demonstrate the efficacy of the numerical method and code.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

First-Principles Modeling Of Electromagnetic Scattering By Discrete and Discretely Heterogeneous Random Media

A discrete random medium is an object in the form of a finite volume of a vacuum or a homogeneous material medium filled with quasi-randomly and quasi-uniformly distributed discrete macroscopic impurities called small particles. Such objects are ubiquitous in natural and artificial environments. They are often characterized by analyzing theoretically the results of laboratory, in situ, or remote-sensing measurements of the scattering of light and other electromagnetic radiation. Electromagnetic scattering and absorption by particles can also affect the energy budget of a discrete random medium and hence various ambient physical and chemical processes. In either case electromagnetic scattering must be modeled in terms of appropriate optical observables, i.e., quadratic or bilinear forms in the field that quantify the reading of a relevant optical instrument or the electromagnetic energy budget. It is generally believed that time-harmonic Maxwell's equations can accurately describe elastic electromagnetic scattering by macroscopic particulate media that change in time much more slowly than the incident electromagnetic field. However, direct solutions of these equations for discrete random media had been impracticable until quite recently. This has led to a widespread use of various phenomenological approaches in situations when their very applicability can be questioned. Recently, however, a new branch of physical optics has emerged wherein electromagnetic scattering by discrete and discretely heterogeneous random media is modeled directly by using analytical or numerically exact computer solutions of the Maxwell equations. Therefore, the main objective of this Report is to formulate the general theoretical framework of electromagnetic scattering by discrete random media rooted in the Maxwell- Lorentz electromagnetics and discuss its immediate analytical and numerical consequences. Starting from the microscopic Maxwell-Lorentz equations, we trace the development of the first principles formalism enabling accurate calculations of monochromatic and quasi-monochromatic scattering by static and randomly varying multiparticle groups. We illustrate how this general framework can be coupled with state-of-the-art computer solvers of the Maxwell equations and applied to direct modeling of electromagnetic scattering by representative random multi-particle groups with arbitrary packing densities. This first-principles modeling yields general physical insights unavailable with phenomenological approaches. We discuss how the first-order-scattering approximation, the radiative transfer theory, and the theory of weak localization of electromagnetic waves can be derived as immediate corollaries of the Maxwell equations for very specific and well-defined kinds of particulate medium. These recent developments confirm the mesoscopic origin of the radiative transfer, weak localization, and effective-medium regimes and help evaluate the numerical accuracy of widely used approximate modeling methodologies.

Michael I Mishchenko↗

Measurement of Transport Properties of Woody Biomass Feedstock Particles Before and After Pyrolysis by Numerical Analysis of X-Ray Tomographic Reconstructions

Lignocellulosic biomass has a complex, species-specific microstructure that governs heat and mass transport during conversion processes. A quantitative understanding of the evolution of pore size and structure is critical to optimize conversion processes for biofuel and bio-based chemical production. Further, improving our understanding of the microstructure of biochar coproduct will accelerate development of its myriad applications. This work quantitatively compares the microstructural features and the anisotropic permeabilities of two woody feedstocks, red oak and Douglas fir, using X-ray computed tomography (XCT) before and after the feedstocks are subjected to pyrolysis. Quantitative analysis of the three-dimensional (3D) reconstructions allows for direct calculations of void fractions, pore size distributions and tortuosity factors. Next, 3D images are imported into an immersed boundary based finite volume solver to simulate gas flow through the porous structure and to directly calculate the principal permeabilities along longitudinal, radial, and tangential directions. The permeabilities of native biomass are seen to differ by three to four orders of magnitude in the different principal directions, but we find that this anisotropy is substantially reduced in the biochar formed during pyrolysis. The quantitative transport properties reported here enhance the ability of pyrolysis simulations to account for feedstock-specific effects and thereby provide a useful touchstone for the biorefining community.

09 BIOMASS FUELS↗

Inverse transport problem solvers based on regularized and compressive sensing techniques

According to the direct exposure measurements from flash radiographic image, regularized-based method and compressive sensing (CS)-based method for inverse transport equation are presented. The linear absorption coefficients and interface locations of objects are reconstructed directly at the same time. With a large number of measurements, least-square method is utilized to complete the reconstruction. Owing to the ill-posedness of the inverse problems, regularized algorithm is employed. Tikhonov method is applied with an appropriate posterior regularization parameter to get a meaningful solution. However, it's always very costly to obtain enough measurements. With limited measurements, CS sparse reconstruction technique Orthogonal Matching Pursuit (OMP) is applied to obtain the sparse coefficients by solving an optimization problem. This paper constructs and takes the forward projection matrix rather than Gauss matrix as measurement matrix. In the CS-based algorithm, Fourier expansion and wavelet expansion are adopted to convert an underdetermined system to a well-posed system. Simulations and numerical results of regularized method with appropriate regularization parameter and that of CS-based agree well with the reference value, furthermore, both methods avoid amplifying the noise. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Preconditioned least‐squares Petrov–Galerkin reduced order models

Abstract In this article, we introduce a methodology for improving the accuracy and efficiency of reduced order models (ROMs) constructed using the least‐squares Petrov–Galerkin (LSPG) projection method through the introduction of preconditioning. Unlike prior related work, which focuses on preconditioning the linear systems arising within the ROM numerical solution procedure to improve linear solver performance, our approach leverages a preconditioning matrix directly within the minimization problem underlying the LSPG formulation. Applying preconditioning in this way has the potential to improve ROM accuracy for several reasons. First, preconditioning the LSPG formulation changes the norm defining the residual minimization, which can improve the residual‐based stability constant bounding the ROM solution's error. The incorporation of a preconditioner into the LSPG formulation can have the additional effect of scaling the components of the residual being minimized to make them roughly of the same magnitude, which can be beneficial when applying the LSPG method to problems with disparate scales (e.g., dimensional equations, multi‐physics problems). Importantly, we demonstrate that an “ideal preconditioned” LSPG ROM (a ROM in which the preconditioner is the inverse of the Jacobian of its corresponding full order model) emulates projection of the full order model solution increment onto the reduced basis. This quantity defines a lower bound on the error of a ROM solution for a given reduced basis. By designing preconditioners that approximate the Jacobian inverse—as is common in designing preconditioners for solving linear systems—it is possible to obtain a ROM whose error approaches this lower bound. The proposed approach is evaluated on several mechanical and thermo‐mechanical problems implemented within the Albany HPC code and run in the predictive regime, with prediction across material parameter space. We demonstrate numerically that the introduction of simple Jacobi, Gauss‐Seidel, and ILU preconditioners into the proper orthogonal decomposition/LSPG formulation reduces significantly the ROM solution error, the reduced Jacobian condition number, the number of nonlinear iterations required to reach convergence, and the wall time (thereby improving efficiency). Moreover, our numerical results reveal that the introduction of preconditioning can deliver a robust and accurate solution for test cases in which the unpreconditioned LSPG method fails to converge.

Lindsay, Payton↗

Scalable implicit incompressible resistive MHD with stabilized FE and fully-coupled Newton–Krylov-AMG

Here, we discuss that the computational solution of the governing balance equations for mass, momentum, heat transfer and magnetic induction for resistive magnetohydrodynamics (MHD) systems can be extremely challenging. These difficulties arise from both the strong nonlinear, nonsymmetric coupling of fluid and electromagnetic phenomena, as well as the significant range of time- and length-scales that the interactions of these physical mechanisms produce. This paper explores the development of a scalable, fully-implicit stabilized unstructured finite element (FE) capability for 3D incompressible resistive MHD. The discussion considers the development of a stabilized FE formulation in context of the variational multiscale (VMS) method, and describes the scalable implicit time integration and direct-to-steady-state solution capability. The nonlinear solver strategy employs Newton–Krylov methods, which are preconditioned using fully-coupled algebraic multilevel preconditioners. These preconditioners are shown to enable a robust, scalable and efficient solution approach for the large-scale sparse linear systems generated by the Newton linearization. Verification results demonstrate the expected order-of-accuracy for the stabilized FE discretization. The approach is tested on a variety of prototype problems, that include MHD duct flows, an unstable hydromagnetic Kelvin–Helmholtz shear layer, and a 3D island coalescence problem used to model magnetic reconnection. Initial results that explore the scaling of the solution methods are also presented on up to 128K processors for problems with up to 1.8B unknowns on a CrayXK7.

97 MATHEMATICS AND COMPUTING↗

Towards a Scalable Fully-Implicit Fully-coupled Resistive MHD Formulation with Stabilized FE Methods

This paper explores the development of a scalable, nonlinear, fully-implicit stabilized unstructured finite element (FE) capability for 2D incompressible (reduced) resistive MHD. The discussion considers the implementation of a stabilized FE formulation in context of a fully-implicit time integration and direct-to-steady-state solution capability. The nonlinear solver strategy employs Newton-Krylov methods, which are preconditioned using fully-coupled algebraic multilevel preconditioners. These preconditioners are shown to enable a robust, scalable and efficient solution approach for the large-scale sparse linear systems generated by the Newton linearization. Verification results demonstrate the expected order-of-accuracy for the stabilized FE discretization. The approach is tested on a variety of prototype problems, including both low-Lundquist number (e.g., an MHD Faraday conduction pump and a hydromagnetic Rayleigh-Bernard linear stability calculation) and moderately-high Lundquist number (magnetic island coalescence problem) examples. Initial results that explore the scaling of the solution methods are presented on up to 4096 processors for problems with up to 64M unknowns on a CrayXT3/4. Additionally, a large-scale proof-of-capability calculation for 1 billion unknowns for the MHD Faraday pump problem on 24,000 cores is presented.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Approximate Green’s Function Coupled Cluster A Method Employing Effective Dimension Reduction

The Green’s function coupled cluster (GFCC) method, originally proposed in the early 1990s, is a powerful many-body tool for computing and analyzing the electronic structures of molecular structures and periodic systems, especially when electrons are strongly correlated. However, in order for the GFCC method to become an approach that can be routinely used in the electronic structure calculations, more robust linear algebraic techniques and approximations need to be employed to reduce its extremely high computational overhead. In our recent studies, we showed that the GFCC equation can be numerically solved by seeking solutions of multiple linear systems in the frequency regime of interest, which can be easily distributed in a massively parallel environment, and broaden the avenue for more state-of-the-art numerical linear algebraic tools associated with multiple linear systems to be kicked in to this field. In the present pursuit of this practice, we show a successful application of model order reduction (MOR) techniques in the GFCC framework. Briefly speaking, for a frequency regime of interest that requires a high resolution of spectral function, instead of solving GFCC linear equation in the original dimension for every single frequency point, an easily-solvable linear system model with reduced dimension can be built upon projecting the original GFCC linear system into a subspace, from which the interpolation and extrapolation can give a reasonable approximation to the results of the original GFCC linear equations. Here, we show that the subspace can be properly constructed in an iterative manner from the auxiliary vectors of the GFCC linear equations at some select frequencies inside the frequency regime. During the iterations, the quality of the subspace, as well as the linear system model, can be systematically improved. The approach is tested in this work in terms of the efficiency and accuracy of computing spectral functions for some typical molecular systems including carbon monoxide, 1,3-butadiene, benzene, and adenine molecules. As a byproduct, the high quality linear system model is also found to be able to provide high quality initial guess for the existing linear solver to improve the convergence performance.

Peng, Bo↗

DESS (Differential Equation System Solver) [SWR-24-48]

The Differential Equation System Solver (DESS) is a Rust crate implementing fixed-step and adaptive-step solvers and designed especially for modeling physical systems. Seven explicit ordinary differential equation (ODE) solver methods have been added so far: Euler’s, Heun’s, Midpoint, Ralston’s, Classic Runge-Kutta, Bogacki-Shampine, and Cash-Karp. These comprise five fixed-step methods and two adaptive-step methods. Few solver packages are implemented in the Rust ecosystem and none are intended specifically for physical system modeling, so the goal of DESS is to create a Rust ODE solver crate designed to easily specify and model physical systems with modular, configurable solver options. In addition to allowing users to directly input equations to solve, DESS allows users to optionally specify and define relationships between nodes in their system, which the package then translates into a system of equations via the Rust macro system, leading to simpler and more intuitive code.

Steuteville, Robin↗

Measuring Input Thresholds on an Existing Board

A critical PECL (positive emitter-coupled logic) interface to Xilinx interface needed to be changed on an existing flight board. The new Xilinx input interface used a CMOS (complementary metal-oxide semiconductor) type of input, and the driver could meet its thresholds typically, but not in worst-case, according to the data sheet. The previous interface had been based on comparison with an external reference, but the CMOS input is based on comparison with an internal divider from the power supply. A way to measure what the exact input threshold was for this device for 64 inputs on a flight board was needed. The measurement technique allowed an accurate measurement of the voltage required to switch a Xilinx input from high to low for each of the 64 lines, while only probing two of them. Directly driving an external voltage was considered too risky, and tests done on any other unit could not be used to qualify the flight board. The two lines directly probed gave an absolute voltage threshold calibration, while data collected on the remaining 62 lines without probing gave relative measurements that could be used to identify any outliers. The PECL interface was forced to a long-period square wave by driving a saturated square wave into the ADC (analog to digital converter). The active pull-down circuit was turned off, causing each line to rise rapidly and fall slowly according to the input s weak pull-down circuitry. The fall time shows up as a change in the pulse width of the signal ready by the Xilinx. This change in pulse width is a function of capacitance, pulldown current, and input threshold. Capacitance was known from the different trace lengths, plus a gate input capacitance, which is the same for all inputs. The pull-down current is the same for all inputs including the two that are probed directly. The data was combined, and the Excel solver tool was used to find input thresholds for the 62 lines. This was repeated over different supply voltages and temperatures to show that the interface had voltage margin under all worst case conditions. Gate input thresholds are normally measured at the manufacturer when the device is on a chip tester. A key function of this machine was duplicated on an existing flight board with no modifications to the nets to be tested, with the exception of changes in the FPGA program.

Kuperman, Igor↗

Technique for Calculating Solution Derivatives With Respect to Geometry Parameters in a CFD Code

A solution has been developed to the challenges of computation of derivatives with respect to geometry, which is not straightforward because these are not typically direct inputs to the computational fluid dynamics (CFD) solver. To overcome these issues, a procedure has been devised that can be used without having access to the mesh generator, while still being applicable to all types of meshes. The basic approach is inspired by the mesh motion algorithms used to deform the interior mesh nodes in a smooth manner when the surface nodes, for example, are in a fluid structure interaction problem. The general idea is to model the mesh edges and nodes as constituting a spring-mass system. Changes to boundary node locations are propagated to interior nodes by allowing them to assume their new equilibrium positions, for instance, one where the forces on each node are in balance. The main advantage of the technique is that it is independent of the volumetric mesh generator, and can be applied to structured, unstructured, single- and multi-block meshes. It essentially reduces the problem down to defining the surface mesh node derivatives with respect to the geometry parameters of interest. For analytical geometries, this is quite straightforward. In the more general case, one would need to be able to interrogate the underlying parametric CAD (computer aided design) model and to evaluate the derivatives either analytically, or by a finite difference technique. Because the technique is based on a partial differential equation (PDE), it is applicable not only to forward mode problems (where derivatives of all the output quantities are computed with respect to a single input), but it could also be extended to the adjoint problem, either by using an analytical adjoint of the PDE or a discrete analog.

Mathur, Sanjay↗

Simulations of a Turbulent Flow Subjected to Favorable and Adverse Pressure Gradients

This paper reports the results from a direct numerical simulation of an initially turbulent boundary layer passing over a wall-mounted “speed bump” geometry. The speed bump, represented in the form of a Gaussian distribution profile, generates a favorable pressure gradient region over the upstream half of the geometry, followed by an adverse pressure gradient over the downstream half. The boundary layer approaching the bump undergoes strong acceleration in the favorable pressure gradient region before experiencing incipient or very weak separation within the adverse pressure gradient region. These types of flows have proven to be particularly challenging to predict using lower-fidelity simulation tools based on various turbulence modeling approaches and warrant the use of the highest-fidelity simulation techniques. Simulation results are utilized to examine the key phenomena present in the flowfield, such as relaminarization/stabilization in the strong acceleration region succeeded by retransition to turbulence near the onset of adverse pressure gradient, incipient/weak separation, and development of internal layers where the sense of streamwise pressure gradient changes at the foot, apex and tail of the bump. The present direct numerical simulation is performed using a flow solver developed exclusively for graphics processing units, which is found to provide a significant speedup compared to an earlier solver optimized for central processing unit architectures.

Ali Uzun↗

Initial performance of fully-coupled AMG and approximate block factorization preconditioners for solution of implicit FE resistive MHD.

This brief paper explores the development of scalable, nonlinear, fully-implicit solution methods for a stabilized unstructured finite element (FE) discretization of the 2D incompressible (reduced) resistive MHD system. The discussion considers the stabilized FE formulation in context of a fully-implicit time integration and direct-to-steady-state solution capability. The nonlinear solver strategy employs Newton-Krylov methods, which are preconditioned using fully-coupled algebraic multilevel (AMG) techniques and a new approximate block factorization (ABF) preconditioner. The intent of these preconditioners is to enable robust, scalable and efficient solution approaches for the large-scale sparse linear systems generated by the Newton linearization. We present results for the fully-coupled AMG preconditioner for two prototype problems, a low Lundquist number MHD Faraday conduction pump and moderately-high Lundquist number incompressible magnetic island coalescence problem. For the MHD pump results we explore the scaling of the fully-coupled AMG preconditioner for up to 4096 processors for problems with up to 64M unknowns on a CrayXT3/4. Using the island coalescence problem we explore the weak scaling of the AMG preconditioner and the influence of the Lundquist number on the iteration count. Finally we present some very recent results for the algorithmic scaling of the ABF preconditioner.

99 GENERAL AND MISCELLANEOUS↗

A stochastic-dynamic model for global atmospheric mass-field statistics

Global atmospheric mass field error correlations based on satellite observations and on numerical forecasts show strong and systematic latitude dependence. A model for the latitude dependent spatial correlation structure of mass field forecast errors is derived from dynamical considerations. Three methods of solution were tested. In the first method, the equation was solved by expansion in spherical harmonics, and the correlation function was computed analytically using the expansion coefficients. In the second method, the finite difference equivalent of the equation was solved using a fast poisson solver. The correlation function was computed using stratified sampling of the individual realizations. In the third method, a higher order equation was derived, and solved directly in finite differences by two successive applications of the fast poisson solver. The three methods were compared for accuracy and efficiency, and the third method was chosen as clearly superior.

Ghil, M.↗

An installed nacelle design method using multiblock Euler solver

An efficient multiblock Euler design method was developed for designing a nacelle installed on geometrically complex airplane configurations. This approach employed a design module based on a direct iterative surface curvature method developed at NASA-Langley. A multiblock Euler flow solver was used for computing flow around complex geometries. The flow solver used a finite-volume formulation with explicit time-stepping to solve the Euler equations. It used a multiblock version of the multigrid method to accelerate the convergence of the calculations. The design module successively updated the surface geometry to reduce the difference between the computed and target pressure distributions. In the flow solver, the change in surface geometry was simulated by applying surface transpiration boundary conditions to avoid repeated grid generation during design iterations. Smoothness of the designed surface was ensured by alternate application of streamwise and circumferential smoothings.

Chen, H. C.↗

Semi-span model testing in the national transonic facility

The present work was motivated by an ongoing research program at NASA Langley Research Center to develop a semi-span testing capability for the National Transonic Facility (NTF). This test technique is being investigated as a means to design and optimize high-lift devices at flight Reynolds numbers in a ground test facility. Even though the freestream Mach numbers of interest are around .20, the flow around a transport wing with high lift devices deployed may contain regions of compressible flow. Thus to properly model the flow physics, a compressible flow solver may be required. However, the application of a compressible flow solver at low Mach numbers can be problematic. The objective of this phase of the project is to directly compare the performance of two widely used three-dimensional compressible Navier-Stokes solvers at low Mach numbers to both experimental data and to results obtained from an incompressible Navier-Stokes solver. The geometries of interest are two isolated wings with different leading edge sweep angles. The compressible Navier-Stokes solvers chosen, TLNS3D-MB and CFL3D, which were developed at NASA Langley Research Center (LaRC), represent the current state-of-the-art in compressible 3-D Navier-Stokes solvers. The incompressible Navier-Stokes solver, INS3D-UP, developed recently at NASA Ames Research Center (ARC), represents the current state-of-the-art in incompressible Navier-Stokes solvers.

Chokani, Ndaona↗

An installed nacelle design code using a multiblock Euler solver. Volume 1: Theory document

An efficient multiblock Euler design code was developed for designing a nacelle installed on geometrically complex airplane configurations. This approach employed a design driver based on a direct iterative surface curvature method developed at LaRC. A general multiblock Euler flow solver was used for computing flow around complex geometries. The flow solver used a finite-volume formulation with explicit time-stepping to solve the Euler Equations. It used a multiblock version of the multigrid method to accelerate the convergence of the calculations. The design driver successively updated the surface geometry to reduce the difference between the computed and target pressure distributions. In the flow solver, the change in surface geometry was simulated by applying surface transpiration boundary conditions to avoid repeated grid generation during design iterations. Smoothness of the designed surface was ensured by alternate application of streamwise and circumferential smoothings. The capability and efficiency of the code was demonstrated through the design of both an isolated nacelle and an installed nacelle at various flow conditions. Information on the execution of the computer program is provided in volume 2.

Chen, H. C.↗

A communication-avoiding 3D sparse triangular solver

We present a novel distributed memory algorithm to improve the strong scalability of the solution of a sparse triangular system. This operation appears in the solve phase of direct methods for solving general sparse linear systems, Ax = b. Our 3D sparse triangular solver employs several techniques, including a 3D MPI process grid, elimination tree parallelism, and data replication, all of which reduce the per-process communication when combined. We present analytical models to understand the communication cost of our algorithm and show that our 3D sparse triangular solver can reduce the per-process communication volume asymptotically by a factor of O(n1/4) and O(n1/6) for problems arising from the finite element discretizations of 2D "planar" and 3D "non-planar" PDEs, respectively. We implement our algorithm for use in SuperLU_DIST3D, using a hybrid MPI+OpenMP programming model. Our 3D triangular solve algorithm, when run on 12k cores of Cray XC30, outperforms the current state-of-the-art 2D algorithm by 7.2x for planar and 2.7x for the non-planar sparse matrices, respectively.

Sao, Piyush↗