Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “algorithm selection”

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 1,117 records · Page 62

Accurate and Scalable O(N) Algorithm for First-Principles Molecular-Dynamics Computations on Large Parallel Computers

We present the first truly scalable first-principles molecular dynamics algorithm with O(N) complexity and controllable accuracy, capable of simulating systems with finite band gaps of sizes that were previously impossible with this degree of accuracy. By avoiding global communications, we provide a practical computational scheme capable of extreme scalability. Accuracy is controlled by the mesh spacing of the finite difference discretization, the size of the localization regions in which the electronic wave functions are confined, and a cutoff beyond which the components of the overlap matrix can be omitted when computing selected elements of its inverse. We demonstrate the algorithm's excellent parallel scaling for up to 101 952 atoms on 23 328 processors, with a wall-clock time of the order of 1 min per molecular dynamics time step and numerical error on the forces of less than 7x10 -4 Ha/Bohr.

36 MATERIALS SCIENCE↗

Mapped Landmark Algorithm for Precision Landing

A report discusses a computer vision algorithm for position estimation to enable precision landing during planetary descent. The Descent Image Motion Estimation System for the Mars Exploration Rovers has been used as a starting point for creating code for precision, terrain-relative navigation during planetary landing. The algorithm is designed to be general because it handles images taken at different scales and resolutions relative to the map, and can produce mapped landmark matches for any planetary terrain of sufficient texture. These matches provide a measurement of horizontal position relative to a known landing site specified on the surface map. Multiple mapped landmarks generated per image allow for automatic detection and elimination of bad matches. Attitude and position can be generated from each image; this image-based attitude measurement can be used by the onboard navigation filter to improve the attitude estimate, which will improve the position estimates. The algorithm uses normalized correlation of grayscale images, producing precise, sub-pixel images. The algorithm has been broken into two sub-algorithms: (1) FFT Map Matching (see figure), which matches a single large template by correlation in the frequency domain, and (2) Mapped Landmark Refinement, which matches many small templates by correlation in the spatial domain. Each relies on feature selection, the homography transform, and 3D image correlation. The algorithm is implemented in C++ and is rated at Technology Readiness Level (TRL) 4.

Johnson, Andrew↗

Orion GN and C Model Based Development: Experience and Lessons Learned

The Orion Guidance Navigation and Control (GN&C) team is charged with developing GN&C algorithms for the Exploration Flight Test One (EFT-1) vehicle. The GN&C team is a joint team consisting primarily of Prime Contractor (Lockheed Martin) and NASA personnel and contractors. Early in the GN&C development cycle the team selected MATLAB/Simulink as the tool for developing GN&C algorithms and Mathworks autocode tools as the means for converting GN&C algorithms to flight software (FSW). This paper provides an assessment of the successes and problems encountered by the GN&C team from the perspective of Orion GN&C developers, integrators, FSW engineers and management. The Orion GN&C approach to graphical development, including simulation tools, standards development and autocode approaches are scored for the main activities that the team has completed through the development phases of the program.

Jackson, Mark C.↗

QSPIN: A High Level Java API for Quantum Computing Experimentation

QSPIN is a high level Java language API for experimentation in QC models used in the calculation of Ising spin glass ground states and related quadratic unconstrained binary optimization (QUBO) problems. The Java API is intended to facilitate research in advanced QC algorithms such as hybrid quantum-classical solvers, automatic selection of constraint and optimization parameters, and techniques for the correction and mitigation of model and solution errors. QSPIN includes high level solver objects tailored to the D-Wave quantum annealing architecture that implement hybrid quantum-classical algorithms [Booth et al.] for solving large problems on small quantum devices, elimination of variables via roof duality, and classical computing optimization methods such as GPU accelerated simulated annealing and tabu search for comparison. A test suite of documented NP-complete applications ranging from graph coloring, covering, and partitioning to integer programming and scheduling are provided to demonstrate current capabilities.

Quantu↗

Time-Dependent Response Versus Scan Angle for MODIS Reflective Solar Bands

The Moderate Resolution Imaging Spectroradiometer (MODIS) instruments currently operate onboard the National Aeronautics and Space Administration (NASA's) Terra and Aqua spacecraft, launched on December 18, 1999 and May 4, 2002, respectively. MODIS has 36 spectral bands, among which 20 are reflective solar bands (RSBs) covering a spectral range from 0.412 to 2.13 μm. The RSBs are calibrated on orbit using a solar diffuser (SD) and an SD stability monitor and with additional measurements from lunar observations via a space view (SV) port. Selected pseudo-invariant desert sites are also used to track the RSB on-orbit gain change, particularly for short-wavelength bands. MODIS views the Earth surface, SV, and the onboard calibrators using a two-sided scan mirror. The response versus scan angle (RVS) of the scan mirror was characterized prior to launch, and its changes are tracked using observations made at different angles of incidence from onboard SD, lunar, and Earth view (EV) measurements. These observations show that the optical properties of the scan mirror have experienced large wavelength-dependent degradation in both the visible and near infrared spectral regions. Algorithms have been developed to track the on-orbit RVS change using the calibrators and the selected desert sites. These algorithms have been applied to both Terra and Aqua MODIS Level 1B (L1B) to improve the EV data accuracy since L1B Collection 4, refined in Collection 5, and further improved in the latest Collection 6 (C6). In C6, two approaches have been used to derive the time-dependent RVS for MODIS RSB. The first approach relies on data collected from sensor onboard calibrators and mirror side ratios from EV observations. The second approach uses onboard calibrators and EV response trending from selected desert sites. This approach is mainly used for the bands with much larger changes in their time-dependent RVS, such as the Terra MODIS bands 1-4, 8, and 9 and the Aqua MODIS bands 8- and 9. In this paper, the algorithms of these approaches are described, their performance is demonstrated, and their impact on L1B products is discussed. In general, the shorter wavelength bands have experienced a larger on-orbit RVS change, which, in general, are mirror side and detector dependent. The on-orbit RVS change due to the degradation of band 8 can be as large as 35 percent for Terra MODIS and 20 percent for Aqua MODIS. Vital to maintaining the accuracy of the MODIS L1B products is an accurate characterization of the on-orbit RVS change. The derived time-independent RVS, implemented in C6, makes an important improvement to the quality of the MODIS L1B products.

MODIS↗

Exploring the contamination of the DES-Y1 cluster sample with SPT-SZ selected clusters

ABSTRACT We perform a cross validation of the cluster catalogue selected by the red-sequence Matched-filter Probabilistic Percolation algorithm (redMaPPer) in Dark Energy Survey year 1 (DES-Y1) data by matching it with the Sunyaev–Zel’dovich effect (SZE) selected cluster catalogue from the South Pole Telescope SPT-SZ survey. Of the 1005 redMaPPer selected clusters with measured richness $\hat{\lambda }\gt 40$ in the joint footprint, 207 are confirmed by SPT-SZ. Using the mass information from the SZE signal, we calibrate the richness–mass relation using a Bayesian cluster population model. We find a mass trend λ ∝ MB consistent with a linear relation (B ∼ 1), no significant redshift evolution and an intrinsic scatter in richness of σλ = 0.22 ± 0.06. By considering two error models, we explore the impact of projection effects on the richness–mass modelling, confirming that such effects are not detectable at the current level of systematic uncertainties. At low richness SPT-SZ confirms fewer redMaPPer clusters than expected. We interpret this richness dependent deficit in confirmed systems as due to the increased presence at low richness of low-mass objects not correctly accounted for by our richness-mass scatter model, which we call contaminants. At a richness $\hat{\lambda }=40$, this population makes up ${\gt}12{{\ \rm per\ cent}}$ (97.5 percentile) of the total population. Extrapolating this to a measured richness $\hat{\lambda }=20$ yields ${\gt}22{{\ \rm per\ cent}}$ (97.5 percentile). With these contamination fractions, the predicted redMaPPer number counts in different plausible cosmologies are compatible with the measured abundance. The presence of such a population is also a plausible explanation for the different mass trends (B ∼ 0.75) obtained from mass calibration using purely optically selected clusters. The mean mass from stacked weak lensing (WL) measurements suggests that these low-mass contaminants are galaxy groups with masses ∼3–5 × 1013 M⊙ which are beyond the sensitivity of current SZE and X-ray surveys but a natural target for SPT-3G and eROSITA.

79 ASTRONOMY AND ASTROPHYSICS↗

Velocity dispersions of clusters in the Dark Energy Survey Y3 redMaPPer catalogue

ABSTRACT We measure the velocity dispersions of clusters of galaxies selected by the red-sequence Matched-filter Probabilistic Percolation (redMaPPer) algorithm in the first three years of data from the Dark Energy Survey (DES), allowing us to probe cluster selection and richness estimation, λ, in light of cluster dynamics. Our sample consists of 126 clusters with sufficient spectroscopy for individual velocity dispersion estimates. We examine the correlations between cluster velocity dispersion, richness, X-ray temperature, and luminosity, as well as central galaxy velocity offsets. The velocity dispersion–richness relation exhibits a bimodal distribution. The majority of clusters follow scaling relations between velocity dispersion, richness, and X-ray properties similar to those found for previous samples; however, there is a significant population of clusters with velocity dispersions that are high for their richness. These clusters account for roughly 22 per cent of the λ < 70 systems in our sample, but more than half (55 per cent) of λ < 70 clusters at z > 0.5. A couple of these systems are hot and X-ray bright as expected for massive clusters with richnesses that appear to have been underestimated, but most appear to have high velocity dispersions for their X-ray properties likely due to line-of-sight structure. These results suggest that projection effects contribute significantly to redMaPPer selection, particularly at higher redshifts and lower richnesses. The redMaPPer determined richnesses for the velocity dispersion outliers are consistent with their X-ray properties, but several are X-ray undetected and deeper data are needed to understand their nature.

79 ASTRONOMY AND ASTROPHYSICS↗

Development of Multi-Sensor Global Cloud and Radiance Composites for DSCOVR EPIC Imager with Subpixel Definition

The Deep Space Climate Observatory (DSCOVR) enables analysis of the daytime Earth radiation budget via the onboard Earth Polychromatic Imaging Camera (EPIC) and National Institute of Standards and Technology Advanced Radiometer (NISTAR). EPIC delivers adequate spatial resolution imagery but only in shortwave bands (317-780 nm), while NISTAR measures the top-of-atmosphere (TOA) whole-disk radiance in shortwave and longwave broadband windows. Accurate calculation of albedo and outgoing longwave flux requires a high-resolution scene identification such as the radiance observations and cloud properties retrievals from low earth orbit (LEO, including NASA Terra and Aqua MODIS, Suomi-NPP VIIRS, and NOAA AVHRR) and geosynchronous (GEO, including GOES east and west, METEOSAT, INSAT-3D, MTSAT-2, and Himawari-8) satellite imagers. The cloud properties are derived using the Clouds and the Earth's Radiant Energy System (CERES) mission Cloud Subsystem group algorithms. These properties have to be co-located with EPIC pixels to provide the scene identification and to select anisotropic directional models (ADMs), which are then used to adjust the NISTAR-measured radiance and subsequently obtain the global daytime shortwave and longwave fluxes. This work presents an algorithm for optimal merging of selected radiance and cloud property parameters derived from multiple satellite imagers to obtain seamless global hourly composites at 5-km resolution. Selection of satellite data for each 5-km pixel is based on an aggregated rating that incorporates five parameters: nominal satellite resolution, pixel time relative to the EPIC time, viewing zenith angle, distance from day/night terminator, and probability of sun glint. To provide a smoother transition in the merged output, in regions where candidate pixel data from two satellite sources have comparable aggregated rating, the selection decision is defined by the cumulative function of the normal distribution so that abrupt changes in the visual appearance of the composite data are avoided. Higher spatial accuracy in the composite product is achieved by using the inverse mapping with gradient search during reprojection and bicubic interpolation for pixel resampling.

Khlopenkov, Konstantin V.↗

Spreader-Bar Radiation Detection System Enhancements: A Modeling and Simulation Study

An investigation of radiation detectors on crane-mounted platforms was performed to understand the limitations and optimal capability of these types of systems. Crane-mounted systems are primarily focused on crane-mounted radiation detector systems, where the detectors are mounted on a spreader-bar platform. Spreader bars are attached with cables to a crane at seaports to load and unload containers from cargo ships. The bars can spread from 20 feet to 53 feet using hydraulics, allowing a single bar to move a variety of container lengths. Spreader-bars are also used on straddle carriers, which move containers around the laydown yard in the port. Spreader bar radiation detection (SBRD) systems are commercially available from at least two companies. These systems employ a set of spectroscopic sensors placed on the underside of a gantry crane and held near the top of a cargo container for detection. These systems have been previously tested and some issues and limitations observed. Radiation detectors mounted on spreader-bars are single sided systems (observing cargo only from the top) and it is therefore challenging to obtain the same sensitivity as double-sided systems (for example, portal monitors) especially when considering the possible amounts of intervening cargo that could be present. The approximately exponential attenuation of gamma rays in intervening dense cargo presents an inherent limitation to detection. However, SBRD systems hold detectors closer to the container, have a longer dwell time, and may have lower background (over the ship); all of these aspects help mitigate the single-sided issue. In addition, there are operational considerations that make spreader-bar radiation systems attractive. These include scanning for radiation during other operations (loading, unloading or container movement) which minimizes additional time for scanning, incorporating the radiation detection into existing equipment to minimize initial cost, and in the case of ship unloading, scanning the containers before reaching the laydown yard. This study is focused on researching optimized configurations of SBRD systems, including number of detectors, locations, and best use of detector data, to determine a reasonable limit on the capability of these types of systems. The results will help support decisions on where and how these systems might be effectively deployed, and could be used for future specification development for SBRD acquisition. The next phase of this research and investigation into SBRD systems and quantification of the possible improved capability with an enhanced SBRD is the focus of this report. Supporting this investigation, detailed models of an enhanced SBRD system were created using a 3D radiation transport code, and models were run to simulate the data from sources and background for a realistic cargo-loading port environment. The enhanced SBRD had increased detector material (numbers of detectors) compared to the commercial systems, with a distribution providing more uniform coverage. Measurements using small industrial sources were made with an enhanced SBRD system developed by Sandia National Laboratory to support model validation to establish credibility for the simulations of these threat and background scenarios. A significant simulation effort was undertaken to simulate sources of interest within a 40-foot cargo container. In order to estimate the performance of a generalized SBRD system in scenarios including background and cargo, a highly enriched uranium (HEU) test source (with 3x10 -8 232 U by weight) within a wood cargo model was simulated for a 40-foot cargo container. This source was placed at a variety of locations, and the flux from the source was multiplied as necessary to create estimated minimal detectable quantities (MDQ) figures for various quantities of HEU. Simulated data sets were produced corresponding to a complete container offloading scenario for a large number of statistical realizations. Significant work was performed on both the algorithm development and the container cargo distributions entering the US, and the result of this modeling, validation, and simulation effort was a set of estimates of the performance of a variety of radiation detection algorithms in this cargo-screening environment. In the case of algorithms, a variety of contemporary radiation detection and identification approaches were employed, and the optimal approaches to selecting time and location subsets of detector data were explored. In all cases, algorithm performance was compared, and the best approaches were suggested along with the best expected performance from the enhanced SBRD. In general, template-matching schemes employing a limited set of regression parameters tend to provide best results. The improvements of an enhanced SBRD with increased detector material and improved algorithms result in an effective MDQ of 10 kg of HEU in ~50 percent of the imported cargo in 40-foot containers, assuming a random source location. This is approximately two times the capability of the estimated performance of the commercial systems. For an MDQ of the 10 kg HEU source located in the worst location (bottom of the 40-foot container), the volume of cargo scanned drops to 10% for the enhanced SBRD, again a factor of two or more larger than the commercial systems. Enhancing the SBRD systems by adding more detectors, providing a more uniform coverage, and incorporating more advanced algorithms will result in a significant increase in capability. Although this investigation provided some indication of the performance capabilities of an enhanced SBRD system, additional work is needed to estimate the capability for other container configurations, such as the 20 and 45 foot containers. Since the 20-foot container weight limit is similar to the 40-foot, more dense cargo is typically shipped in 20-foot containers. This will result in a smaller percentage of cargo scanned for the 20-foot containers (for a particular source quantity) compared to the 40-foot containers. In addition to other configuration, other source material needs to be investigated to understand the capabilities for the entire threat space, where only HEU is considered in this work.

crane-mounted↗

Hybrid Inverse Parameter Identification of Fully Coupled Ductile Damage Model for Steel Sheet DP600 with Two Different Algorithms: Trust Region and Genetic Algorithms

Ductile damage model has been widely regarded as a valuable method to predict the failure of sheet metal. Based on the thermodynamic theory and continuum damage mechanics, the fully coupled ductile damage model can be developed, which also can better predict the initiation and growth of the fracture. But the identifications of model parameters with theoretical methodology are difficult due to the complex coupling relationships existing among all state variables. The inverse methodology is regarded as a good method to resolve the problem. In this paper, the recently proposed fully coupled ductile damage model is chosen to investigate the deformation behavior of DP600 steel, in which the mixed saturation isotropic and kinematic hardenings are taken into account and fully coupled with the ductile damage. During the identification process, the least square formula of the error between experimental and numerical results is selected as the target function. The trust region algorithm and genetic algorithm are used with the help of MATLAB for the identification of three damage parameters. Finally, by comparing the experimental and numerical results, the validations of two algorithms are proved. The efficiency of the optimization process with trust region algorithm is higher, but with lower accuracy. Meanwhile, the optimization process is greatly affected by the chosen initial values of the ductile damage parameters.

36 MATERIALS SCIENCE↗

Antenna analysis using neural networks

Conventional computing schemes have long been used to analyze problems in electromagnetics (EM). The vast majority of EM applications require computationally intensive algorithms involving numerical integration and solutions to large systems of equations. The feasibility of using neural network computing algorithms for antenna analysis is investigated. The ultimate goal is to use a trained neural network algorithm to reduce the computational demands of existing reflector surface error compensation techniques. Neural networks are computational algorithms based on neurobiological systems. Neural nets consist of massively parallel interconnected nonlinear computational elements. They are often employed in pattern recognition and image processing problems. Recently, neural network analysis has been applied in the electromagnetics area for the design of frequency selective surfaces and beam forming networks. The backpropagation training algorithm was employed to simulate classical antenna array synthesis techniques. The Woodward-Lawson (W-L) and Dolph-Chebyshev (D-C) array pattern synthesis techniques were used to train the neural network. The inputs to the network were samples of the desired synthesis pattern. The outputs are the array element excitations required to synthesize the desired pattern. Once trained, the network is used to simulate the W-L or D-C techniques. Various sector patterns and cosecant-type patterns (27 total) generated using W-L synthesis were used to train the network. Desired pattern samples were then fed to the neural network. The outputs of the network were the simulated W-L excitations. A 20 element linear array was used. There were 41 input pattern samples with 40 output excitations (20 real parts, 20 imaginary). A comparison between the simulated and actual W-L techniques is shown for a triangular-shaped pattern. Dolph-Chebyshev is a different class of synthesis technique in that D-C is used for side lobe control as opposed to pattern shaping. The interesting thing about D-C synthesis is that the side lobes have the same amplitude. Five-element arrays were used. Again, 41 pattern samples were used for the input. Nine actual D-C patterns ranging from -10 dB to -30 dB side lobe levels were used to train the network. A comparison between simulated and actual D-C techniques for a pattern with -22 dB side lobe level is shown. The goal for this research was to evaluate the performance of neural network computing with antennas. Future applications will employ the backpropagation training algorithm to drastically reduce the computational complexity involved in performing EM compensation for surface errors in large space reflector antennas.

Smith, William T.↗

Space Launch System Implementation of Adaptive Augmenting Control

Given the complex structural dynamics, challenging ascent performance requirements, and rigorous flight certification constraints owing to its manned capability, the NASA Space Launch System (SLS) launch vehicle requires a proven thrust vector control algorithm design with highly optimized parameters to provide stable and high-performance flight. On its development path to Preliminary Design Review (PDR), the SLS flight control system has been challenged by significant vehicle flexibility, aerodynamics, and sloshing propellant. While the design has been able to meet all robust stability criteria, it has done so with little excess margin. Through significant development work, an Adaptive Augmenting Control (AAC) algorithm has been shown to extend the envelope of failures and flight anomalies the SLS control system can accommodate while maintaining a direct link to flight control stability criteria such as classical gain and phase margin. In this paper, the work performed to mature the AAC algorithm as a baseline component of the SLS flight control system is presented. The progress to date has brought the algorithm design to the PDR level of maturity. The algorithm has been extended to augment the full SLS digital 3-axis autopilot, including existing load-relief elements, and the necessary steps for integration with the production flight software prototype have been implemented. Several updates which have been made to the adaptive algorithm to increase its performance, decrease its sensitivity to expected external commands, and safeguard against limitations in the digital implementation are discussed with illustrating results. Monte Carlo simulations and selected stressing case results are also shown to demonstrate the algorithm's ability to increase the robustness of the integrated SLS flight control system.

Wall, John H.↗

A Marker-Based Approach for the Automated Selection of a Single Segmentation from a Hierarchical Set of Image Segmentations

The Hierarchical SEGmentation (HSEG) algorithm, which combines region object finding with region object clustering, has given good performances for multi- and hyperspectral image analysis. This technique produces at its output a hierarchical set of image segmentations. The automated selection of a single segmentation level is often necessary. We propose and investigate the use of automatically selected markers for this purpose. In this paper, a novel Marker-based HSEG (M-HSEG) method for spectral-spatial classification of hyperspectral images is proposed. Two classification-based approaches for automatic marker selection are adapted and compared for this purpose. Then, a novel constrained marker-based HSEG algorithm is applied, resulting in a spectral-spatial classification map. Three different implementations of the M-HSEG method are proposed and their performances in terms of classification accuracies are compared. The experimental results, presented for three hyperspectral airborne images, demonstrate that the proposed approach yields accurate segmentation and classification maps, and thus is attractive for remote sensing image analysis.

Tarabalka, Y.↗

On current compositing algorithms

Several techniques exist for compositing the multitemporal NOAA Advanced Very High Resolution Radiometer (AVHRR) data for vegetation studies. The major pixel selection criteria of these techniques rely on the characteristics of the NDVI (Normalized Difference Vegetative Index): appearance of clouds, poor atmospheric conditions, and off-nadir viewing geometries would depress the NDVI values. Consequently, selecting the pixels with the maximum value of NDVI would presumably eliminate these external perturbating effects. However, the maximum NDVI does not always correspond to these ideal conditions. In fact, the NDVI varies with these external factors in an unpredictable way. There was an indication that the maximum NDVI tended to favor the off-nadir view in the forward direction. The resultant composite product would be consequently affected. To improve the multitemporal data via compositing, therefore, both the pixel selection criteria and the classifier NDVI need to be modified or corrected for external factors. The current compositing algorithms were reviewed, and alternatives were proposed to use the combinations of the red and near infrared channels and biological characteristics of vegetation as second criteria in pixel selections. The traditional classifier NDVI was replaced with different vegetation indices. The approach was applied to an AVHRR data set over the HAPEX study site in Niger in 1992. The results showed that the approach improved the AVHRR time series quality and was promising towards the development of an efficient compositing algorithm.

Qi, J.↗

Study report on interfacing major physiological subsystem models: An approach for developing a whole-body algorithm

Using a whole body algorithm simulation model, a wide variety and large number of stresses as well as different stress levels were simulated including environmental disturbances, metabolic changes, and special experimental situations. Simulation of short term stresses resulted in simultaneous and integrated responses from the cardiovascular, respiratory, and thermoregulatory subsystems and the accuracy of a large number of responding variables was verified. The capability of simulating significantly longer responses was demonstrated by validating a four week bed rest study. In this case, the long term subsystem model was found to reproduce many experimentally observed changes in circulatory dynamics, body fluid-electrolyte regulation, and renal function. The value of systems analysis and the selected design approach for developing a whole body algorithm was demonstrated.

Fitzjerrell, D. G.↗

Algorithms for airborne Doppler radar wind shear detection

Honeywell has developed algorithms for the detection of wind shear/microburst using airborne Doppler radar. The Honeywell algorithms use three dimensional pattern recognition techniques and the selection of an associated scanning pattern forward of the aircraft. This 'volumetric scan' approach acquires reflectivity, velocity, and spectral width from a three dimensional volume as opposed to the conventional use of a two dimensional azimuthal slice of data at a fixed elevation. The algorithm approach is based on detection and classification of velocity patterns which are indicative of microburst phenomenon while minimizing the false alarms due to ground clutter return. Simulation studies of microburst phenomenon and x-band radar interaction with the microburst have been performed and results of that study are presented. Algorithm performance indetection of both 'wet' and 'dry' microbursts is presented.

Gillberg, Jeff↗

Latent uncertainties of the precalculated track Monte Carlo method

Purpose: While significant progress has been made in speeding up Monte Carlo (MC) dose calculation methods, they remain too time-consuming for the purpose of inverse planning. To achieve clinically usable calculation speeds, a precalculated Monte Carlo (PMC) algorithm for proton and electron transport was developed to run on graphics processing units (GPUs). The algorithm utilizes pregenerated particle track data from conventional MC codes for different materials such as water, bone, and lung to produce dose distributions in voxelized phantoms. While PMC methods have been described in the past, an explicit quantification of the latent uncertainty arising from the limited number of unique tracks in the pregenerated track bank is missing from the paper. With a proper uncertainty analysis, an optimal number of tracks in the pregenerated track bank can be selected for a desired dose calculation uncertainty. Methods: Particle tracks were pregenerated for electrons and protons using EGSnrc and GEANT4 and saved in a database. The PMC algorithm for track selection, rotation, and transport was implemented on the Compute Unified Device Architecture (CUDA) 4.0 programming framework. PMC dose distributions were calculated in a variety of media and compared to benchmark dose distributions simulated from the corresponding general-purpose MC codes in the same conditions. A latent uncertainty metric was defined and analysis was performed by varying the pregenerated track bank size and the number of simulated primary particle histories and comparing dose values to a “ground truth” benchmark dose distribution calculated to 0.04% average uncertainty in voxels with dose greater than 20% of D{sub max}. Efficiency metrics were calculated against benchmark MC codes on a single CPU core with no variance reduction. Results: Dose distributions generated using PMC and benchmark MC codes were compared and found to be within 2% of each other in voxels with dose values greater than 20% of the maximum dose. In proton calculations, a small (≤1 mm) distance-to-agreement error was observed at the Bragg peak. Latent uncertainty was characterized for electrons and found to follow a Poisson distribution with the number of unique tracks per energy. A track bank of 12 energies and 60000 unique tracks per pregenerated energy in water had a size of 2.4 GB and achieved a latent uncertainty of approximately 1% at an optimal efficiency gain over DOSXYZnrc. Larger track banks produced a lower latent uncertainty at the cost of increased memory consumption. Using an NVIDIA GTX 590, efficiency analysis showed a 807 × efficiency increase over DOSXYZnrc for 16 MeV electrons in water and 508 × for 16 MeV electrons in bone. Conclusions: The PMC method can calculate dose distributions for electrons and protons to a statistical uncertainty of 1% with a large efficiency gain over conventional MC codes. Before performing clinical dose calculations, models to calculate dose contributions from uncharged particles must be implemented. Following the successful implementation of these models, the PMC method will be evaluated as a candidate for inverse planning of modulated electron radiation therapy and scanned proton beams.

07 ISOTOPE AND RADIATION SOURCES↗