Search NASA⌕ Search

SEARCH · Search NASA

Results for “kernel method”

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 793 records · Page 44

Steps toward fault-tolerant quantum chemistry.

Developing quantum chemistry programs on the coming generation of exascale computers will be a difficult task. The programs will need to be fault-tolerant and minimize the use of global operations. This work explores the use a task-based model that uses a data-centric approach to allocate work to different processes as it applies to quantum chemistry. After introducing the key problems that appear when trying to parallelize a complicated quantum chemistry method such as coupled-cluster theory, we discuss the implications of that model as it pertains to the computational kernel of a coupled-cluster program - matrix multiplication. Also, we discuss the extensions that would required to build a full coupled-cluster program using the task-based model. Current programming models for high-performance computing are fault-intolerant and use global operations. Those properties are unsustainable as computers scale to millions of CPUs; instead one must recognize that these systems will be hierarchical in structure, prone to constant faults, and global operations will be infeasible. The FAST-OS HARE project is introducing a scale-free computing model to address these issues. This model is hierarchical and fault-tolerant by design, allows for the clean overlap of computation and communication, reducing the network load, does not require checkpointing, and avoids the complexity of many HPC runtimes. Development of an algorithm within this model requires a change in focus from imperative programming to a data-centric approach. Quantum chemistry (QC) algorithms, in particular electronic structure methods, are an ideal test bed for this computing model. These methods describe the distribution of electrons in a molecule, which determine the properties of the molecule. The computational cost of these methods is high, scaling quartically or higher in the size of the molecule, which is why QC applications are major users of HPC resources. The complexity of these algorithms means that MPI alone is insufficient to achieve parallel scaling; QC developers have been forced to use alternative approaches to achieve scalability and would be receptive to radical shifts in the programming paradigm. Initial work in adapting the simplest QC method, Hartree-Fock, to this the new programming model indicates that the approach is beneficial for QC applications. However, the advantages to being able to scale to exascale computers are greatest for the computationally most expensive algorithms; within QC these are the high-accuracy coupled-cluster (CC) methods. Parallel coupledcluster programs are available, however they are based on the conventional MPI paradigm. Much of the effort is spent handling the complicated data dependencies between the various processors, especially as the size of the problem becomes large. The current paradigm will not survive the move to exascale computers. Here we discuss the initial steps toward designing and implementing a CC method within this model. First, we introduce the general concepts behind a CC method, focusing on the aspects that make these methods difficult to parallelize with conventional techniques. Then we outline what is the computational core of the CC method - a matrix multiply - within the task-based approach that the FAST-OS project is designed to take advantage of. Finally we outline the general setup to implement the simplest CC method in this model, linearized CC doubles (LinCC).

99 GENERAL AND MISCELLANEOUS↗

Theoretical detection threshold of the proton-acoustic range verification technique

Purpose: Range verification in proton therapy using the proton-acoustic signal induced in the Bragg peak was investigated for typical clinical scenarios. The signal generation and detection processes were simulated in order to determine the signal-to-noise limits. Methods: An analytical model was used to calculate the dose distribution and local pressure rise (per proton) for beams of different energy (100 and 160 MeV) and spot widths (1, 5, and 10 mm) in a water phantom. In this method, the acoustic waves propagating from the Bragg peak were generated by the general 3D pressure wave equation implemented using a finite element method. Various beam pulse widths (0.1–10 μs) were simulated by convolving the acoustic waves with Gaussian kernels. A realistic PZT ultrasound transducer (5 cm diameter) was simulated with a Butterworth bandpass filter with consideration of random noise based on a model of thermal noise in the transducer. The signal-to-noise ratio on a per-proton basis was calculated, determining the minimum number of protons required to generate a detectable pulse. The maximum spatial resolution of the proton-acoustic imaging modality was also estimated from the signal spectrum. Results: The calculated noise in the transducer was 12–28 mPa, depending on the transducer central frequency (70–380 kHz). The minimum number of protons detectable by the technique was on the order of 3–30 × 10{sup 6} per pulse, with 30–800 mGy dose per pulse at the Bragg peak. Wider pulses produced signal with lower acoustic frequencies, with 10 μs pulses producing signals with frequency less than 100 kHz. Conclusions: The proton-acoustic process was simulated using a realistic model and the minimal detection limit was established for proton-acoustic range validation. These limits correspond to a best case scenario with a single large detector with no losses and detector thermal noise as the sensitivity limiting factor. Our study indicated practical proton-acoustic range verification may be feasible with approximately 5 × 10{sup 6} protons/pulse and beam current.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

A new method for diagnosing effective radiative forcing from aerosol–cloud interactions in climate models

Aerosol–cloud interactions (ACIs) are a leading source of uncertainty in estimates of the historical effective radiative forcing (ERF). One reason for this uncertainty is the difficulty in estimating the ERF from aerosol–cloud interactions (ERFaci) in climate models, which typically requires multiple calls to the radiation code. Most commonly used methods also cannot disentangle the contributions from different processes to ERFaci. Here, we develop a new, computationally efficient method for estimating the shortwave (SW) ERFaci from liquid clouds using histograms of monthly averaged cloud fraction partitioned by cloud droplet effective radius (r e ) and liquid water path (LWP). Multiplying the histograms with SW cloud radiative kernels gives the total SW ERFaci from liquid clouds, which can be decomposed into contributions from the Twomey effect, LWP adjustments, and cloud fraction (CF) adjustments. We test the method with data from five CMIP6-era models, using the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite instrument simulator to generate the histograms. Our method gives similar total SW ERFaci estimates to other established methods in regions of prevalent liquid cloud and indicates that the Twomey effect, LWP adjustments, and CF adjustments have contributed -0.34 ± 0.23, -0.22 ± 0.13, and -0.09 ± 0.11 W m -2 , respectively, to the effective radiative forcing of the climate since 1850 in the ensemble mean (95 % confidence). These results demonstrate that widespread adoption of a MODIS r e –LWP joint histogram diagnostic would allow the SW ERFaci and its components to be quickly and accurately diagnosed from climate model outputs, a crucial step for reducing uncertainty in the historical ERF.

54 ENVIRONMENTAL SCIENCES↗

MO-F-CAMPUS-J-02: Automatic Recognition of Patient Treatment Site in Portal Images Using Machine Learning

Purpose: To investigate the method to automatically recognize the treatment site in the X-Ray portal images. It could be useful to detect potential treatment errors, and to provide guidance to sequential tasks, e.g. automatically verify the patient daily setup. Methods: The portal images were exported from MOSAIQ as DICOM files, and were 1) processed with a threshold based intensity transformation algorithm to enhance contrast, and 2) where then down-sampled (from 1024×768 to 128×96) by using bi-cubic interpolation algorithm. An appearance-based vector space model (VSM) was used to rearrange the images into vectors. A principal component analysis (PCA) method was used to reduce the vector dimensions. A multi-class support vector machine (SVM), with radial basis function kernel, was used to build the treatment site recognition models. These models were then used to recognize the treatment sites in the portal image. Portal images of 120 patients were included in the study. The images were selected to cover six treatment sites: brain, head and neck, breast, lung, abdomen and pelvis. Each site had images of the twenty patients. Cross-validation experiments were performed to evaluate the performance. Results: MATLAB image processing Toolbox and scikit-learn (a machine learning library in python) were used to implement the proposed method. The average accuracies using the AP and RT images separately were 95% and 94% respectively. The average accuracy using AP and RT images together was 98%. Computation time was ∼0.16 seconds per patient with AP or RT image, ∼0.33 seconds per patient with both of AP and RT images. Conclusion: The proposed method of treatment site recognition is efficient and accurate. It is not sensitive to the differences of image intensity, size and positions of patients in the portal images. It could be useful for the patient safety assurance. The work was partially supported by a research grant from Varian Medical System.

60 APPLIED LIFE SCIENCES↗

Stochastic, real-space, imaginary-time evaluation of third-order Feynman–Goldstone diagrams

A new, alternative set of interpretation rules of Feynman–Goldstone diagrams for many-body perturbation theory is proposed, which translates diagrams into algebraic expressions suitable for direct Monte Carlo integrations. A vertex of a diagram is associated with a Coulomb interaction (rather than a two-electron integral) and an edge with the trace of a Green's function in real space and imaginary time. With these, 12 diagrams of third-order many-body perturbation (MP3) theory are converted into 20-dimensional integrals, which are then evaluated by a Monte Carlo method. It uses redundant walkers for convergence acceleration and a weight function for importance sampling in conjunction with the Metropolis algorithm. The resulting Monte Carlo MP3 method has low-rank polynomial size dependence of the operation cost, a negligible memory cost, and a naturally parallel computational kernel, while reproducing the correct correlation energies of small molecules within a few mE{sub h} after 10{sup 6} Monte Carlo steps.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Overdestabilization vs Overstabilization in the Theoretical Analysis of f-Orbital Covalency

The complex nature of the f-orbital electronic structures and their interaction with the chemical environment pose significant computational challenges. Advanced computational techniques that variationally include scalar relativities and spin–orbit coupling directly at the molecular orbital level have been developed to address this complexity. Among these, variational relativistic multiconfigurational multireference methods stand out for their high accuracy and systematic improvement in studies of f-block complexes. Additionally, these advanced methods offer the potential for calibrating low-scaling electronic structure methods such as density functional theory. However, studies on the Cl K-edge X-ray absorption spectra of the [Ce(III)Cl 6 ] 3– and [Ce(IV)Cl 6 ] 2– complexes show that time-dependent density functional theory with approximate exchange–correlation kernels can lead to inaccuracies, resulting in an overstabilization of 4f orbitals and incorrect assessments of covalency. In contrast, approaches utilizing small active space wave function methods may understate the stability of these orbitals. The results herein demonstrate the need for large active space, multireference, and variational relativistic methods in studying f-block complexes.

chemical calculations↗

Reconstruction of the primordial power spectrum of curvature perturbations using multiple data sets

Detailed knowledge of the primordial power spectrum of curvature perturbations is essential both in order to elucidate the physical mechanism ('inflation') which generated it, and for estimating the cosmological parameters from observations of the cosmic microwave background and large-scale structure. Hence it ought to be extracted from such data in a model-independent manner, however this is difficult because relevant cosmological observables are given by a convolution of the primordial perturbations with some smoothing kernel which depends on both the assumed world model and the matter content of the universe. Moreover the deconvolution problem is ill-conditioned so a regularisation scheme must be employed to control error propagation. We demonstrate that 'Tikhonov regularisation' can robustly reconstruct the primordial spectrum from multiple cosmological data sets, a significant advantage being that both its uncertainty and resolution are then quantified. Using Monte Carlo simulations we investigate several regularisation parameter selection methods and find that generalised cross-validation and Mallow's C{sub p} method give optimal results. We apply our inversion procedure to data from the Wilkinson Microwave Anisotropy Probe, other ground-based small angular scale CMB experiments, and the Sloan Digital Sky Survey. The reconstructed spectrum (assuming the standard ΛCDM cosmology) is not scale-free but has an infrared cutoff at k∼<5 × 10{sup −4} Mpc{sup −1} (due to the anomalously low CMB quadrupole) and several features with ∼ 2σ significance at k/Mpc{sup −1} ∼ 0.0013–0.0025, 0.0362–0.0402 and 0.051–0.056, reflecting the 'WMAP glitches'. To test whether these are indeed real will require more accurate data, such as from the Planck satellite and new ground-based experiments.

79 ASTRONOMY AND ASTROPHYSICS↗

An integrated manifold learning approach for high-dimensional data feature extractions and its applications to online process monitoring of additive manufacturing

As an effective dimension reduction and feature extraction technique, manifold learning has been successfully applied to high-dimensional data analysis. With the rapid development of sensor technology, a large amount of high-dimensional data such as image streams can be easily available. Thus, a promising application of manifold learning is in the field of sensor signal analysis, particular for the applications of online process monitoring and control using high-dimensional data. The objective of this study is to develop a manifold learning-based feature extraction method for process monitoring of Additive Manufacturing (AM) using online sensor data. Due to the non-parametric nature of most existing manifold learning methods, their performance in terms of computational efficiency, as well as noise resistance has yet to be improved. To address this issue, this study proposes an integrated manifold learning approach termed multi-kernel metric learning embedded isometric feature mapping (MKML-ISOMAP) for dimension reduction and feature extraction of online high-dimensional sensor data such as images. Based on the extracted features with the utilization of supervised classification and regression methods, an online process monitoring methodology for AM is implemented to identify the actual process quality status. Finally, in the numerical simulation and real-world case studies, the proposed method demonstrates excellent performance in both prediction accuracy and computational efficiency.

36 MATERIALS SCIENCE↗

A robust and efficient approach to detect 3D rectal tubes from CT colonography

Purpose: The rectal tube (RT) is a common source of false positives (FPs) in computer-aided detection (CAD) systems for CT colonography. A robust and efficient detection of RT can improve CAD performance by eliminating such ''obvious'' FPs and increase radiologists' confidence in CAD. Methods: In this paper, we present a novel and robust bottom-up approach to detect the RT. Probabilistic models, trained using kernel density estimation on simple low-level features, are employed to rank and select the most likely RT tube candidate on each axial slice. Then, a shape model, robustly estimated using random sample consensus (RANSAC), infers the global RT path from the selected local detections. Subimages around the RT path are projected into a subspace formed from training subimages of the RT. A quadratic discriminant analysis (QDA) provides a classification of a subimage as RT or non-RT based on the projection. Finally, a bottom-top clustering method is proposed to merge the classification predictions together to locate the tip position of the RT. Results: Our method is validated using a diverse database, including data from five hospitals. On a testing data with 21 patients (42 volumes), 99.5% of annotated RT paths have been successfully detected. Evaluated with CAD, 98.4% of FPs caused by the RT have been detected and removed without any loss of sensitivity. Conclusions: The proposed method demonstrates a high detection rate of the RT path, and when tested in a CAD system, reduces FPs caused by the RT without the loss of sensitivity.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Pre-Ionization Controlled Laser Plasma Formation for Ignition Applications (Final Scientific/Technical Report)

This report consists of three grand funded papers. Abstracts for each follow. 1) In this work, we present experimental and modeling studies of air pre-ionization using ultraviolet (UV) laser pulses and its effect on laser breakdown of an overlapped near-infrared (NIR) pulse. Experimental studies are conducted with a 266 nm beam (fourth harmonic of Nd:YAG) for UV pre-ionization and an overlapped 1064 nm NIR beam (fundamental of Nd:YAG), both having pulse duration of 10 ns. Results show that the UV beam produces a pre-ionized volume which assists in breakdown of the NIR beam, leading to reduction in NIR breakdown threshold by factor of >2. Numerical modeling is performed to examine the ionization and breakdown of both beams. The modeled breakdown threshold of the NIR, including assist by pre-ionization, is in reasonable agreement with the experimental results. 2) The present contribution compares the energy absorption, optical emission, temperature, and fluid dynamics of ultraviolet (UV) λ = 266 nm and near infrared (NIR) λ = 1064 nm nanosecond laser induced plasmas in ambient air. For UV pulses at the conditions studied, energy absorption by the plasmas increases relatively gradually with laser pulse energy starting at delivered energy of E~8 mJ. Corresponding measurements of plasma luminosity show that the absorption of UV radiation does not necessarily result in visible plasma emission. For the NIR induced plasmas, the energy absorption profile is far more abrupt and begins at ~55 mJ. In contrast with UV, the absorption of NIR radiation is always accompanied by intense optical emission. The temperatures of both types of plasma have been measured with Rayleigh scattering thermometry (at times after the Thomson signal sufficiently diminishes). The UV plasmas can attain a wider range of temperatures, including lower temperatures, depending on the pulse energy (e.g., T ~400–2000K for E ~7–35 mJ at Δt = 10 μs after the pulse) while the NIR plasmas show only hotter temperatures (e.g., T ~ 12 000 K for E = 75 mJ at Δt = 10 μs after the pulse) as is consistent with the literature. Differences in the fluid dynamics for UV versus NIR pulses are shown with Schlieren imaging. In this work, the contrast in the UV and NIR plasma threshold behavior is attributed to differing roles of avalanche ionization and multiphoton ionization as is also illustrated by a simple numerical model. 3) The present contribution examines the impact of plasma dynamics and plasma-driven fluid dynamics on the flame growth of laser ignited mixtures and shows that a new dual-pulse scheme can be used to control the kernel formation process in ways that extend the lean ignition limit. We perform a comparative study between (conventional) single-pulse laser ignition (λ = 1064 nm) and a novel dual-pulse method based on combining an ultraviolet (UV) pre-ionization pulse (λ = 266 nm) with an overlapped near-infrared (NIR) energy addition pulse (λ = 1064 nm). We employ OH* chemiluminescence to visualize the evolution of the early flame kernel. For single-pulse laser ignition at lean conditions, the flame kernel separates through third lobe detachment, corresponding to high strain rates that extinguish the flame. In this work, we investigate the capabilities of the dual-pulse to control the plasma-driven fluid dynamics by adjusting the axial offset of the two focal points. In particular, we find there exists a beam waist offset whereby the resulting vorticity suppresses formation of the third lobe, consequently reducing flame stretch. With this approach, we demonstrate that the dual-pulse method enables reduced flame speeds (at early times), an extended lean limit, increased combustion efficiency, and decreased laser energy requirements.

Mechanical engineering↗

EXAGRAPH: Graph and combinatorial methods for enabling exascale applications

Combinatorial algorithms in general and graph algorithms in particular play a critical enabling role in numerous scientific applications. However, the irregular memory access nature of these algorithms makes them one of the hardest algorithmic kernels to implement on parallel systems. With tens of billions of hardware threads and deep memory hierarchies, the exascale computing systems in particular pose extreme challenges in scaling graph algorithms. The codesign center on combinatorial algorithms, ExaGraph, was established to design and develop methods and techniques for efficient implementation of key combinatorial (graph) algorithms chosen from a diverse set of exascale applications. Algebraic and combinatorial methods have a complementary role in the advancement of computational science and engineering, including playing an enabling role on each other. In this paper, we survey the algorithmic and software development activities performed under the auspices of ExaGraph from both a combinatorial and an algebraic perspective. In particular, we detail our recent efforts in porting the algorithms to manycore accelerator (GPU) architectures. We also provide a brief survey of the applications that have benefited from the scalable implementations of different combinatorial algorithms to enable scientific discovery at scale. We believe that several applications will benefit from the algorithmic and software tools developed by the ExaGraph team.

97 MATHEMATICS AND COMPUTING↗

An Immersed Boundary method with divergence-free velocity interpolation and force spreading

The Immersed Boundary (IB) method is a mathematical framework for constructing robust numerical methods to study fluid–structure interaction in problems involving an elastic structure immersed in a viscous fluid. The IB formulation uses an Eulerian representation of the fluid and a Lagrangian representation of the structure. The Lagrangian and Eulerian frames are coupled by integral transforms with delta function kernels. The discretized IB equations use approximations to these transforms with regularized delta function kernels to interpolate the fluid velocity to the structure, and to spread structural forces to the fluid. It is well-known that the conventional IB method can suffer from poor volume conservation since the interpolated Lagrangian velocity field is not generally divergence-free, and so this can cause spurious volume changes. In practice, the lack of volume conservation is especially pronounced for cases where there are large pressure differences across thin structural boundaries. The aim of this paper is to greatly reduce the volume error of the IB method by introducing velocity-interpolation and force-spreading schemes with the properties that the interpolated velocity field in which the structure moves is at least $\mathscr{C}$ 1 and satisfies a continuous divergence-free condition, and that the force-spreading operator is the adjoint of the velocity-interpolation operator. We confirm through numerical experiments in two and three spatial dimensions that this new IB method is able to achieve substantial improvement in volume conservation compared to other existing IB methods, at the expense of a modest increase in the computational cost. Further, the new method provides smoother Lagrangian forces (tractions) than traditional IB methods. Finally, the method presented here is restricted to periodic computational domains. Its generalization to non-periodic domains is important future work.

97 MATHEMATICS AND COMPUTING↗

Variational Approach to Yang-Mills Theory with non-Gaussian Wave Functionals

A general method for treating non-Gaussian wave functionals in quantum field theory is presented and applied to the Hamiltonian approach to Yang-Mills theory in Coulomb gauge in order to include a three-gluon kernel in the exponential of the vacuum wave functional. The three-gluon vertex is calculated using the propagators found in the variational approach with a Gaussian trial wave functional as input.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Automated detection of pulmonary nodules in PET/CT images: Ensemble false-positive reduction using a convolutional neural network technique

Purpose: Automated detection of solitary pulmonary nodules using positron emission tomography (PET) and computed tomography (CT) images shows good sensitivity; however, it is difficult to detect nodules in contact with normal organs, and additional efforts are needed so that the number of false positives (FPs) can be further reduced. In this paper, the authors propose an improved FP-reduction method for the detection of pulmonary nodules in PET/CT images by means of convolutional neural networks (CNNs). Methods: The overall scheme detects pulmonary nodules using both CT and PET images. In the CT images, a massive region is first detected using an active contour filter, which is a type of contrast enhancement filter that has a deformable kernel shape. Subsequently, high-uptake regions detected by the PET images are merged with the regions detected by the CT images. FP candidates are eliminated using an ensemble method; it consists of two feature extractions, one by shape/metabolic feature analysis and the other by a CNN, followed by a two-step classifier, one step being rule based and the other being based on support vector machines. Results: The authors evaluated the detection performance using 104 PET/CT images collected by a cancer-screening program. The sensitivity in detecting candidates at an initial stage was 97.2%, with 72.8 FPs/case. After performing the proposed FP-reduction method, the sensitivity of detection was 90.1%, with 4.9 FPs/case; the proposed method eliminated approximately half the FPs existing in the previous study. Conclusions: An improved FP-reduction scheme using CNN technique has been developed for the detection of pulmonary nodules in PET/CT images. The authors’ ensemble FP-reduction method eliminated 93% of the FPs; their proposed method using CNN technique eliminates approximately half the FPs existing in the previous study. These results indicate that their method may be useful in the computer-aided detection of pulmonary nodules using PET/CT images.

60 APPLIED LIFE SCIENCES↗

A dose point kernel database using GATE Monte Carlo simulation toolkit for nuclear medicine applications: Comparison with other Monte Carlo codes

Purpose: GATE is a Monte Carlo simulation toolkit based on the Geant4 package, widely used for many medical physics applications, including SPECT and PET image simulation and more recently CT image simulation and patient dosimetry. The purpose of the current study was to calculate dose point kernels (DPKs) using GATE, compare them against reference data, and finally produce a complete dataset of the total DPKs for the most commonly used radionuclides in nuclear medicine. Methods: Patient-specific absorbed dose calculations can be carried out using Monte Carlo simulations. The latest version of GATE extends its applications to Radiotherapy and Dosimetry. Comparison of the proposed method for the generation of DPKs was performed for (a) monoenergetic electron sources, with energies ranging from 10 keV to 10 MeV, (b) beta emitting isotopes, e.g., {sup 177}Lu, {sup 90}Y, and {sup 32}P, and (c) gamma emitting isotopes, e.g., {sup 111}In, {sup 131}I, {sup 125}I, and {sup 99m}Tc. Point isotropic sources were simulated at the center of a sphere phantom, and the absorbed dose was stored in concentric spherical shells around the source. Evaluation was performed with already published studies for different Monte Carlo codes namely MCNP, EGS, FLUKA, ETRAN, GEPTS, and PENELOPE. A complete dataset of total DPKs was generated for water (equivalent to soft tissue), bone, and lung. This dataset takes into account all the major components of radiation interactions for the selected isotopes, including the absorbed dose from emitted electrons, photons, and all secondary particles generated from the electromagnetic interactions. Results: GATE comparison provided reliable results in all cases (monoenergetic electrons, beta emitting isotopes, and photon emitting isotopes). The observed differences between GATE and other codes are less than 10% and comparable to the discrepancies observed among other packages. The produced DPKs are in very good agreement with the already published data, which allowed us to produce a unique DPKs dataset using GATE. The dataset contains the total DPKs for {sup 67}Ga, {sup 68}Ga, {sup 90}Y, {sup 99m}Tc, {sup 111}In, {sup 123}I, {sup 124}I, {sup 125}I, {sup 131}I, {sup 153}Sm, {sup 177}Lu {sup 186}Re, and {sup 188}Re generated in water, bone, and lung. Conclusions: In this study, the authors have checked GATE's reliability for absorbed dose calculation when transporting different kind of particles, which indicates its robustness for dosimetry applications. A novel dataset of DPKs is provided, which can be applied in patient-specific dosimetry using analytical point kernel convolution algorithms.

61 RADIATION PROTECTION AND DOSIMETRY↗

End-to-end GPU acceleration of low-order-refined preconditioning for high-order finite element discretizations

In this article, we present algorithms and implementations for the end-to-end GPU acceleration of matrix-free low-order-refined preconditioning of high-order finite element problems. The methods described here allow for the construction of effective preconditioners for high-order problems with optimal memory usage and computational complexity. The preconditioners are based on the construction of a spectrally equivalent low-order discretization on a refined mesh, which is then amenable to, for example, algebraic multigrid preconditioning. The constants of equivalence are independent of mesh size and polynomial degree. For vector finite element problems in H(curl) and H(div) (e.g., for electromagnetic or radiation diffusion problems), a specially constructed interpolation–histopolation basis is used to ensure fast convergence. Detailed performance studies are carried out to analyze the efficiency of the GPU algorithms. The kernel throughput of each of the main algorithmic components is measured, and the strong and weak parallel scalability of the methods is demonstrated. The different relative weighting and significance of the algorithmic components on GPUs and CPUs is discussed. Results on problems involving adaptively refined nonconforming meshes are shown, and the use of the preconditioners on a large-scale magnetic diffusion problem using all spaces of the finite element de Rham complex is illustrated.

97 MATHEMATICS AND COMPUTING↗

On comparing helioseismic two-dimensional inversion methods

We consider inversion techniques for investigating the structure and dynamics of the solar interior as functions of radius and latitude. In particular, we look at the problem of inferring the radial and latitudinal dependence of the Sun's internal rotation, using a fully two-dimensional least-squares inversion algorithm. Concepts such as averaging kernels, measures of resolution, and trade-off curves, which have previously been used in the one-dimensional case, are generalized to facilitate a comparison of two-dimensional methods. We investigate the weighting given to different modes and discuss the implications of this for observational strategies. As an illustration we use a mode set whose properties are similar to those expected for data from the GONG network.

Schou, J.↗

High Energy Electromagnetic Particle Transportation on the GPU

We present massively parallel high energy electromagnetic particle transportation through a finely segmented detector on a Graphics Processing Unit (GPU). Simulating events of energetic particle decay in a general-purpose high energy physics (HEP) detector requires intensive computing resources, due to the complexity of the geometry as well as physics processes applied to particles copiously produced by primary collisions and secondary interactions. The recent advent of hardware architectures of many-core or accelerated processors provides the variety of concurrent programming models applicable not only for the high performance parallel computing, but also for the conventional computing intensive application such as the HEP detector simulation. The components of our prototype are a transportation process under a non-uniform magnetic field, geometry navigation with a set of solid shapes and materials, electromagnetic physics processes for electrons and photons, and an interface to a framework that dispatches bundles of tracks in a highly vectorized manner optimizing for spatial locality and throughput. Core algorithms and methods are excerpted from the Geant4 toolkit, and are modified and optimized for the GPU application. Program kernels written in C/C++ are designed to be compatible with CUDA and OpenCL and with the aim to be generic enough for easy porting to future programming models and hardware architectures. To improve throughput by overlapping data transfers with kernel execution, multiple CUDA streams are used. Issues with floating point accuracy, random numbers generation, data structure, kernel divergences and register spills are also considered. Performance evaluation for the relative speedup compared to the corresponding sequential execution on CPU is presented as well.

Canal, P.↗