Search NASA⌕ Search

SEARCH · Search NASA

Results for “tensor processing units”

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

Adaptive variational quantum dynamics simulations with compressed circuits and fewer measurements

The adaptive variational quantum dynamics simulation (AVQDS) method performs real-time evolution of quantum states using automatically generated parametrized quantum circuits that often contain substantially fewer gates than Trotter circuits. Here we report an improved version of the method, which we call AVQDS(T), by porting the tiling efficient trial circuits with rotations implemented simultaneously technique. The algorithm adaptively adds layers of disjoint unitary gates to the ansatz circuit so as to keep the McLachlan distance, a measure of the accuracy of the variational dynamics, below a fixed threshold. Here we perform benchmark noiseless AVQDS(T) simulations of quench dynamics in local spin models and compare with an alternative adaptive variational approach on quantum resource requirement. Quantum dynamical simulations implementing realistic noise channels are also reported. Finally, we propose a way to substantially alleviate the measurement overhead of AVQDS(T) while maintaining high accuracy by synergistically integrating quantum circuit calculations on quantum processing units with classical calculations using, e.g., tensor networks to evaluate the quantum geometric tensor. We showcase that this approach enables AVQDS(T) to deliver more accurate results than simulations using a fixed ansatz of comparable final depth for a significant time duration with fewer quantum resources.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

On the Feasibility of Using Reduced-Precision Tensor Core Operations for Graph Analytics

Today’s data-driven analytics and machine learning workload have been largely driven by the General-PurposeGraphics Processing Units (GPGPUs). To accelerate dense matrix multiplications on the GPUs, Tensor Core Units (TCUs) have been introduced in recent years. In this paper, we study linear-algebra-based and vertex-centric algorithms for various graph kernels on the GPUs with an objective of applying this new hardware feature to graph applications. We identify the potential stages in these graph kernels that can be executed on the Tensor Core Units. In particular, we leverage the reformulation of the reduction and scan operations in terms of matrix multiplication [1]on the TCUs. We demonstrate that executing these operations on the TCUs, available inside different graph kernels, can assist in establishing an end-to-end pipeline on the GPGPUs without depending on hand-tuned external libraries and still can deliver comparable performance for various graph analytics.

Graph algorithms, GPU computing↗

GPU-acceleration of tensor renormalization with PyTorch using CUDA

We show that numerical computations based on tensor renormalization group (TRG) methods can be significantly accelerated with PyTorch on graphics processing units (GPUs) by leveraging NVIDIA's Compute Unified Device Architecture (CUDA). Here we find improvement in the runtime and its scaling with bond dimension for two-dimensional systems. Our results establish that the utilization of GPU resources is essential for future precision computations with TRG.

97 MATHEMATICS AND COMPUTING↗

GPAW: An open Python package for electronic structure calculations

We review the GPAW open-source Python package for electronic structure calculations. GPAW is based on the projector-augmented wave method and can solve the self-consistent density functional theory (DFT) equations using three different wave-function representations, namely real-space grids, plane waves, and numerical atomic orbitals. The three representations are complementary and mutually independent and can be connected by transformations via the real-space grid. This multi-basis feature renders GPAW highly versatile and unique among similar codes. By virtue of its modular structure, the GPAW code constitutes an ideal platform for the implementation of new features and methodologies. Moreover, it is well integrated with the Atomic Simulation Environment (ASE), providing a flexible and dynamic user interface. In addition to ground-state DFT calculations, GPAW supports many-body GW band structures, optical excitations from the Bethe–Salpeter Equation, variational calculations of excited states in molecules and solids via direct optimization, and real-time propagation of the Kohn–Sham equations within time-dependent DFT. A range of more advanced methods to describe magnetic excitations and non-collinear magnetism in solids are also now available. In addition, GPAW can calculate non-linear optical tensors of solids, charged crystal point defects, and much more. Recently, support for graphics processing unit (GPU) acceleration has been achieved with minor modifications to the GPAW code thanks to the CuPy library. We end the review with an outlook, describing some future plans for GPAW.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Diamagnetic studies on as-processed carbon fibers

The Faraday method has been used to measure the diamagnetic susceptibilities of small bundles of aligned carbon fibers of different types at room temperature. It was found that the tensor trace susceptibility of the fibers varies systematically over the range 0.8-20 (in units of -10 to the -6th emu/g) as a function of precursor type and processing history. The susceptibility increases, in general, with increasing nominal treatment temperature and hot stretching, and with increasing tensile elastic modulus. The anisotropy ratio of fibers increases approximately linearly with tensile elastic modulus for all fibers from about 1 for a modulus of about 70 GN/sq m to about 22 for a modulus of 700 GN/sq m in air, and is quantitatively consistent with the layer-plane orientation textures determined by X-ray diffraction when appropriate values of the crystallite principal susceptibilities are used.

Scott, C. B.↗

A sequential DNN based Baseline Energy Prediction Framework with Long term Error Mitigation

In this paper,we aim to present a novel sequential framework containing two deep network architectures for baseline energy prediction, applied to a simulated commercial building dataset. We have observed that this sequential framework helps to improve long term prediction accuracy, there by mitigating prediction error accumulation over time. The proposed framework utilizes convolution layers to extract features from the input data space without changing spatial relations between variables. These features are memorized by tensor train based gated recurrent units for an accurate long-term prediction. Although, architecture comprising the amalgamation of convolution layers and memory cell have shown promising results in domains such as social media analysis, language modeling, video frame prediction and image recognition, this paper extends its scope to the context of energy applications. Furthermore, the addition of tensor train based gated recurrent units is motivated by the necessity of computational time reduction during the training process, there by making this framework more suitable for future in-field deployment. Lastly, results show the current framework outperforms other existing machine learning based methods, in both short-term and long-term prediction categories.

Chakraborty, Indrasis↗

Leapfrog Time-Stepping for Hermite Methods

We introduce here Hermite-leapfrog methods for first order linear wave systems. The new Hermite-leapfrog methods pair leapfrog time-stepping with the Hermite methods of Goodrich and co-authors et al. (Math Comput 75(254):595–630, 2006). The new schemes stagger field variables in both time and space and are high-order accurate for equations with smooth solutions and coefficients. We provide a detailed description of the method and demonstrate that the method conserves variable quantities. Higher dimensional versions of the method are constructed via tensor products. Numerical evidence and rigorous analysis in one space dimension establish stability and high-order convergence. Experiments demonstrating efficient implementations on a graphics processing unit are also presented.

97 MATHEMATICS AND COMPUTING↗

Accelerating Scientific Computing in the Post-Moore’s Era

Novel uses of graphical processing units for accelerated computation revolutionized the field of high-performance scientific computing by providing specialized workflows tailored to algorithmic requirements. As the era of Moore’s law draws to a close, many new non–von Neumann processors are emerging as potential computational accelerators, including those based on the principles of neuromorphic computing, tensor algebra, and quantum information. While development of these new processors is continuing to mature, the potential impact on accelerated computing is anticipated to be profound. We discuss how different processing models can advance computing in key scientific paradigms: machine learning and constraint satisfaction. Significantly, each of these new processor types utilizes a fundamentally different model of computation, and this raises questions about how to best use such processors in the design and implementation of applications. While many processors are being developed with a specific domain target, the ubiquity of spin-glass models and neural networks provides an avenue for multi-functional applications. Furthermore, this also hints at the infrastructure needed to integrate next-generation processing units into future high-performance computing systems.

97 MATHEMATICS AND COMPUTING↗

Towards Precision-Aware Fault Tolerance Approaches for Mixed-Precision Applications

Graphics Processing Units (GPUs), the dominantly adopted accelerators in HPC systems, are susceptible to transient hardware fault. New generation of GPUs feature mixed-precision architectures such as NVIDIA Tensor Cores to accelerate matrix multiplications. While being widely adapted, how would they behave under transient hardware faults remain unclear. In this study, we conduct a large-scale fault injection experiments on GEMM kernels implemented with different floating-point data types on the V100 and A100 Tensor Cores, and show distinct error resilience characteristics for the GEMMS with different formats. In the future, we plan to explore this space by building precision-aware floating-point fault tolerance techniques for applications such as DNNs that exercise low-precision computations.

Fang, Bo↗

Hydrodynamics of suspensions of passive and active rigid particles: a rigid multiblob approach

We develop a rigid multiblob method for numerically solving the mobility problem for suspensions of passive and active rigid particles of complex shape in Stokes flow in unconfined, partially confined, and fully confined geometries. As in a number of existing methods, we discretize rigid bodies using a collection of minimally resolved spherical blobs constrained to move as a rigid body, to arrive at a potentially large linear system of equations for the unknown Lagrange multipliers and rigid-body motions. Here we develop a block-diagonal preconditioner for this linear system and show that a standard Krylov solver converges in a modest number of iterations that is essentially independent of the number of particles. Key to the efficiency of the method is a technique for fast computation of the product of the blob-blob mobility matrix and a vector. For unbounded suspensions, we rely on existing analytical expressions for the Rotne-Prager-Yamakawa tensor combined with a fast multipole method (FMM) to obtain linear scaling in the number of particles. For suspensions sedimented against a single no-slip boundary, we use a direct summation on a graphical processing unit (GPU), which gives quadratic asymptotic scaling with the number of particles. For fully confined domains, such as periodic suspensions or suspensions confined in slit and square channels, we extend a recently developed rigid-body immersed boundary method by B. Kallemov, A. P. S. Bhalla, B. E. Griffith, and A. Donev (Commun. Appl. Math. Comput. Sci. 11 (2016), no. 1, 79-141) to suspensions of freely moving passive or active rigid particles at zero Reynolds number. We demonstrate that the iterative solver for the coupled fluid and rigid-body equations converges in a bounded number of iterations regardless of the system size. In our approach, each iteration only requires a few cycles of a geometric multigrid solver for the Poisson equation, and an application of the block-diagonal preconditioner, leading to linear scaling with the number of particles. We optimize a number of parameters in the iterative solvers and apply our method to a variety of benchmark problems to carefully assess the accuracy of the rigid multiblob approach as a function of the resolution. We also model the dynamics of colloidal particles studied in recent experiments, such as passive boomerangs in a slit channel, as well as a pair of non-Brownian active nanorods sedimented against a wall.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Atomic orbital-based SOS-MP2 with tensor hypercontraction. I. GPU-based tensor construction and exploiting sparsity

We present a tensor hypercontracted (THC) scaled opposite spin second order Møller-Plesset perturbation theory (SOS-MP2) method. By using THC, we reduce the formal scaling of SOS-MP2 with respect to molecular size from quartic to cubic. We achieve further efficiency by exploiting sparsity in the atomic orbitals and using graphical processing units (GPUs) to accelerate integral construction and matrix multiplication. The practical scaling of GPU-accelerated atomic orbital-based THC-SOS-MP2 calculations is found to be N{sup 2.6} for reference data sets of water clusters and alanine polypeptides containing up to 1600 basis functions. The errors in correlation energy with respect to density-fitting-SOS-MP2 are less than 0.5 kcal/mol for all systems tested (up to 162 atoms).

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Reduced scaling formulation of CASPT2 analytical gradients using the supporting subspace method

We present a reduced scaling and exact reformulation of state specific complete active space second-order perturbation (CASPT2) analytical gradients in terms of the MP2 and Fock derivatives using the supporting subspace method. This work follows naturally from the supporting subspace formulation of the CASPT2 energy in terms of the MP2 energy using dressed orbitals and Fock builds. For a given active space configuration, the terms corresponding to the MP2-gradient can be evaluated with O(N5) operations, while the rest of the calculations can be computed with O(N3) operations using Fock builds, Fock gradients, and linear algebra. When tensor-hyper-contraction is applied simultaneously, the computational cost can be further reduced to O(N4) for a fixed active space size. The new formulation enables efficient implementation of CASPT2 analytical gradients by leveraging the existing graphical processing unit (GPU)-based MP2 and Fock routines. We present benchmark results that demonstrate the accuracy and performance of the new method. Example applications of the new method in ab initio molecular dynamics simulation and constrained geometry optimization are given.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Propagation Pattern for Moment Representation of the Lattice Boltzmann Method

A propagation pattern for the moment representation of the regularized lattice Boltzmann method (LBM) in three dimensions is presented. Using effectively lossless compression, the simulation state is stored as a set of moments of the lattice Boltzmann distribution function, instead of the distribution function itself. An efficient cache-aware propagation pattern for this moment representation has the effect of substantially reducing both the storage and memory bandwidth required for LBM simulations. This article extends recent work with the moment representation by expanding the performance analysis on central processing unit (CPU) architectures, considering how boundary conditions are implemented, and demonstrating the effectiveness of the moment representation on a graphics processing unit (GPU) architecture.

42 ENGINEERING↗

TAMM: Tensor algebra for many-body methods

Tensor algebra operations such as contractions in computational chemistry consume a significant fraction of the computing time on large-scale computing platforms. The widespread use of tensor contractions between large multi-dimensional tensors in describing electronic structure theory has motivated the development of multiple tensor algebra frameworks targeting heterogeneous computing platforms. In this paper, we present Tensor Algebra for Many-body Methods (TAMM), a framework for productive and performance-portable development of scalable computational chemistry methods. TAMM decouples the specification of the computation from the execution of these operations on available high-performance computing systems. With this design choice, the scientific application developers (domain scientists) can focus on the algorithmic requirements using the tensor algebra interface provided by TAMM, whereas high-performance computing developers can direct their attention to various optimizations on the underlying constructs, such as efficient data distribution, optimized scheduling algorithms, and efficient use of intra-node resources (e.g., graphics processing units). The modular structure of TAMM allows it to support different hardware architectures and incorporate new algorithmic advances. We describe the TAMM framework and our approach to the sustainable development of scalable ground- and excited-state electronic structure methods. We present case studies highlighting the ease of use, including the performance and productivity gains compared to other frameworks.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

NbF{sub 5} and TaF{sub 5}: Assignment of {sup 19}F NMR resonances and chemical bond analysis from GIPAW calculations

The {sup 19}F isotropic chemical shifts (δ{sub iso}) of two isomorphic compounds, NbF{sub 5} and TaF{sub 5}, which involve six nonequivalent fluorine sites, have been experimentally determined from the reconstruction of 1D {sup 19}F MAS NMR spectra. In parallel, the corresponding {sup 19}F chemical shielding tensors have been calculated using the GIPAW method for both experimental and DFT-optimized structures. Furthermore, the [M{sub 4}F{sub 20}] units of NbF{sub 5} and TaF{sub 5} being held together by van der Waals interactions, the relevance of Grimme corrections to the DFT optimization processes has been evaluated. However, the semi-empirical dispersion correction term introduced by such a method does not show any significant improvement. Nonetheless, a complete and convincing assignment of the {sup 19}F NMR lines of NbF{sub 5} and TaF{sub 5} is obtained, ensured by the linearity between experimental {sup 19}F δ{sub iso} values and calculated {sup 19}F isotropic chemical shielding σ{sub iso} values. The effects of the geometry optimizations have been carefully analyzed, confirming among other matters, the inaccuracy of the experimental structure of NbF{sub 5}. The relationships between the fluorine chemical shifts, the nature of the fluorine atoms (bridging or terminal), the position of the terminal ones (opposite or perpendicular to the bridging ones), the fluorine charges, the ionicity and the length of the M–F bonds have been established. Additionally, for three of the {sup 19}F NMR lines of NbF{sub 5}, distorted multiplets, arising from {sup 1}J-coupling and residual dipolar coupling between the {sup 19}F and {sup 93}Nb nuclei, were simulated yielding to values of {sup 93}Nb–{sup 19}F {sup 1}J-coupling for the corresponding fluorine sites. - Graphical abstract: The complete assignment of the {sup 19}F NMR lines of NbF{sub 5} and TaF{sub 5} allow establishing relationships between the {sup 19}F δ{sub iso} values, the nature of the fluorine atoms (bridging or terminal), the position of the terminal ones (opposite or perpendicular to the bridging ones), the fluorine charges, the ionicity and the length of the M–F bonds. Display Omitted - Highlights: • The {sup 19}F δ{sub iso} values of NbF{sub 5} and TaF{sub 5} have been determined. • The {sup 19}F chemical shielding tensors have been calculated using the GIPAW method. • A confident assignment of the {sup 19}F NMR lines of NbF{sub 5} and TaF{sub 5} is obtained. • The relationships between the {sup 19}Fδ{sub iso} values and the M–F bonds features are established.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

A self-documenting source-independent data format for computer processing of tensor time series

The UCLA Space Science Group has developed a fixed format intermediate data set called a block data set, which is designed to hold multiple segments of multicomponent sampled data series. The format is sufficiently general so that tensor functions of one or more independent variables can be stored in the form of virtual data. This makes it possible for the unit data records of the block data set to be arrays of a single dependent variable rather than discrete samples. The format is self-documenting with parameter, label and header records completely characterizing the contents of the file. The block data set has been applied to the filing of satellite data (of ATS-6 among others).

Mcpherron, R. L.↗

Accurate numerical simulations of open quantum systems using spectral tensor trains

Decoherence between qubits is a major bottleneck in quantum computations. Decoherence results from intrinsic quantum and thermal fluctuations as well as noise in the external fields that perform the measurement and preparation processes. With prescribed colored noise spectra for intrinsic and extrinsic noise, we present a numerical method, Quantum Accelerated Stochastic Propagator Evaluation (Q-ASPEN), to solve the time-dependent noise-averaged reduced density matrix in the presence of intrinsic and extrinsic noise. Q-ASPEN is arbitrarily accurate and can be applied to provide estimates for the resources needed to error-correct quantum computations. We employ spectral tensor trains, which combine the advantages of tensor networks and pseudospectral methods, as a variational ansatz to the quantum relaxation problem and optimize the ansatz using methods typically used to train neural networks. Here, the spectral tensor trains in Q-ASPEN make accurate calculations with tens of quantum levels feasible. We present benchmarks for Q-ASPEN on the spin-boson model in the presence of intrinsic noise and on a quantum chain of up to 32 sites in the presence of extrinsic noise. In our benchmark, the memory cost of Q-ASPEN scales as a low-order polynomial in the size of the system once the number of system states surpasses the number of basis functions used in the spectral expansion.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗