Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “CARLO”

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 199 records · Page 11

TH-E-18A-01: Developments in Monte Carlo Methods for Medical Imaging

Monte Carlo simulation methods are widely used in medical physics research and are starting to be implemented in clinical applications such as radiation therapy planning systems. Monte Carlo simulations offer the capability to accurately estimate quantities of interest that are challenging to measure experimentally while taking into account the realistic anatomy of an individual patient. Traditionally, practical application of Monte Carlo simulation codes in diagnostic imaging was limited by the need for large computational resources or long execution times. However, recent advancements in high-performance computing hardware, combined with a new generation of Monte Carlo simulation algorithms and novel postprocessing methods, are allowing for the computation of relevant imaging parameters of interest such as patient organ doses and scatter-to-primaryratios in radiographic projections in just a few seconds using affordable computational resources. Programmable Graphics Processing Units (GPUs), for example, provide a convenient, affordable platform for parallelized Monte Carlo executions that yield simulation times on the order of 10{sup 7} xray/ s. Even with GPU acceleration, however, Monte Carlo simulation times can be prohibitive for routine clinical practice. To reduce simulation times further, variance reduction techniques can be used to alter the probabilistic models underlying the x-ray tracking process, resulting in lower variance in the results without biasing the estimates. Other complementary strategies for further reductions in computation time are denoising of the Monte Carlo estimates and estimating (scoring) the quantity of interest at a sparse set of sampling locations (e.g. at a small number of detector pixels in a scatter simulation) followed by interpolation. Beyond reduction of the computational resources required for performing Monte Carlo simulations in medical imaging, the use of accurate representations of patient anatomy is crucial to the virtual generation of medical images and accurate estimation of radiation dose and other imaging parameters. For this, detailed computational phantoms of the patient anatomy must be utilized and implemented within the radiation transport code. Computational phantoms presently come in one of three format types, and in one of four morphometric categories. Format types include stylized (mathematical equation-based), voxel (segmented CT/MR images), and hybrid (NURBS and polygon mesh surfaces). Morphometric categories include reference (small library of phantoms by age at 50th height/weight percentile), patient-dependent (larger library of phantoms at various combinations of height/weight percentiles), patient-sculpted (phantoms altered to match the patient's unique outer body contour), and finally, patient-specific (an exact representation of the patient with respect to both body contour and internal anatomy). The existence and availability of these phantoms represents a very important advance for the simulation of realistic medical imaging applications using Monte Carlo methods. New Monte Carlo simulation codes need to be thoroughly validated before they can be used to perform novel research. Ideally, the validation process would involve comparison of results with those of an experimental measurement, but accurate replication of experimental conditions can be very challenging. It is very common to validate new Monte Carlo simulations by replicating previously published simulation results of similar experiments. This process, however, is commonly problematic due to the lack of sufficient information in the published reports of previous work so as to be able to replicate the simulation in detail. To aid in this process, the AAPM Task Group 195 prepared a report in which six different imaging research experiments commonly performed using Monte Carlo simulations are described and their results provided. The simulation conditions of all six cases are provided in full detail, with all necessary data on material composition, source, geometry, scoring and other parameters provided. The results of these simulations when performed with the four most common publicly available Monte Carlo packages are also provided in tabular form. The Task Group 195 Report will be useful for researchers needing to validate their Monte Carlo work, and for trainees needing to learn Monte Carlo simulation methods. In this symposium we will review the recent advancements in highperformance computing hardware enabling the reduction in computational resources needed for Monte Carlo simulations in medical imaging. We will review variance reduction techniques commonly applied in Monte Carlo simulations of medical imaging systems and present implementation strategies for efficient combination of these techniques with GPU acceleration. Trade-offs involved in Monte Carlo acceleration by means of denoising and “sparse sampling” will be discussed. A method for rapid scatter correction in cone-beam CT (<5 min/scan) will be presented as an illustration of the simulation speeds achievable with optimized Monte Carlo simulations. We will also discuss the development, availability, and capability of the various combinations of computational phantoms for Monte Carlo simulation of medical imaging systems. Finally, we will review some examples of experimental validation of Monte Carlo simulations and will present the AAPM Task Group 195 Report. Learning Objectives: Describe the advances in hardware available for performing Monte Carlo simulations in high performance computing environments. Explain variance reduction, denoising and sparse sampling techniques available for reduction of computational time needed for Monte Carlo simulations of medical imaging. List and compare the computational anthropomorphic phantoms currently available for more accurate assessment of medical imaging parameters in Monte Carlo simulations. Describe experimental methods used for validation of Monte Carlo simulations in medical imaging. Describe the AAPM Task Group 195 Report and its use for validation and teaching of Monte Carlo simulations in medical imaging.

60 APPLIED LIFE SCIENCES↗

gRASPA

GPU Monte Carlo Simulation Code with a taste of RASPA We present enhancements in Monte Carlo simulation speed and functionality within an open-source code, gRASPA, which uses graphical processing units (GPUs) to achieve significant performance improvements compared to serial, CPU implementations of Monte Carlo. The code supports a wide range of Monte Carlo simulations, including canonical ensemble (NVT), grand canonical, NVT Gibbs, Widom test particle insertions, and continuous-fractional component Monte Carlo. Implementation of grand canonical transition matrix Monte Carlo (GC-TMMC) and a novel feature to allow different moves for the different components of metal-organic framework (MOF) structures exemplify the capabilities of gRASPA for precise free energy calculations and enhanced adsorption studies, respectively. The introduction of a High-Throughput Computing (HTC) mode permits many Monte Carlo simulations on a single GPU device for accelerated materials discovery. The code can incorporate machine learning (ML) potentials. The open-source nature of gRASPA promotes reproducibility and openness in science, and users may add features to the code and optimize it for their own purposes. The code is written in CUDA/C++ and SYCL/C++ to support different GPU vendors. The gRASPA code is publicly available at https://github.com/snurr-group/gRASPA.

Li, Zhao [Purdue/Northwestern/Notre Dame Universit↗

On the use of Monte Carlo-derived dosimetric data in the estimation of patient dose from CT examinations

The purpose of this work was to investigate the applicability and appropriateness of Monte Carlo-derived normalized data to provide accurate estimations of patient dose from computed tomography (CT) exposures. Monte Carlo methodology and mathematical anthropomorphic phantoms were used to simulate standard patient CT examinations of the head, thorax, abdomen, and trunk performed on a multislice CT scanner. Phantoms were generated to simulate the average adult individual and two individuals with different body sizes. Normalized dose values for all radiosensitive organs and normalized effective dose values were calculated for standard axial and spiral CT examinations. Discrepancies in CT dosimetry using Monte Carlo-derived coefficients originating from the use of: (a) Conversion coefficients derived for axial CT exposures, (b) a mathematical anthropomorphic phantom of standard body size to derive conversion coefficients, and (c) data derived for a specific CT scanner to estimate patient dose from CT examinations performed on a different scanner, were separately evaluated. The percentage differences between the normalized organ dose values derived for contiguous axial scans and the corresponding values derived for spiral scans with pitch=1 and the same total scanning length were up to 10%, while the corresponding percentage differences in normalized effective dose values were less than 0.7% for all standard CT examinations. The normalized organ dose values for standard spiral CT examinations with pitch 0.5-1.5 were found to differ from the corresponding values derived for contiguous axial scans divided by the pitch, by less than 14% while the corresponding percentage differences in normalized effective dose values were less than 1% for all standard CT examinations. Normalized effective dose values for the standard contiguous axial CT examinations derived by Monte Carlo simulation were found to considerably decrease with increasing body size of the mathematical phantom used. When the body-mass index was increased from 23.0 to 32.7 kg/m{sup 2} discrepancies in patient effective dose were up to 34%. The error in estimating effective dose from a CT exposure performed on a specific CT scanner using Monte Carlo data derived for a different CT scanner was estimated to be up to 25%. A simple method was proposed and validated for the determination of scanner-specific normalized dosimetric data from data derived from Monte Carlo simulation of a specific scanner. In conclusion, computed tomography dose index (CTDI) to effective dose conversion coefficients derived by Monte Carlo simulation of axial CT scans may provide a good approximation of corresponding coefficients applicable in helical scans. However, the use of Monte Carlo conversion coefficients for the estimation of patient dose from a CT examination involves a remarkable inaccuracy when the body size of the mathematical anthropomorphic phantom used in Monte Carlo simulation differs from the body of the patient. Therefore, separate sets of Monte Carlo dosimetric CT data shall be generated for different patient body sizes. Besides calculation of different sets of Monte Carlo data for each commercially available scanner is not necessary, since scanner specific data may be derived with acceptable accuracy from the Monte Carlo data calculated for a specific scanner appropriately modified for the different CTDI{sub w}/CTDI{sub air} ratio.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Monte Carlo reference data sets for imaging research: Executive summary of the report of AAPM Research Committee Task Group 195

The use of Monte Carlo simulations in diagnostic medical imaging research is widespread due to its flexibility and ability to estimate quantities that are challenging to measure empirically. However, any new Monte Carlo simulation code needs to be validated before it can be used reliably. The type and degree of validation required depends on the goals of the research project, but, typically, such validation involves either comparison of simulation results to physical measurements or to previously published results obtained with established Monte Carlo codes. The former is complicated due to nuances of experimental conditions and uncertainty, while the latter is challenging due to typical graphical presentation and lack of simulation details in previous publications. In addition, entering the field of Monte Carlo simulations in general involves a steep learning curve. It is not a simple task to learn how to program and interpret a Monte Carlo simulation, even when using one of the publicly available code packages. This Task Group report provides a common reference for benchmarking Monte Carlo simulations across a range of Monte Carlo codes and simulation scenarios. In the report, all simulation conditions are provided for six different Monte Carlo simulation cases that involve common x‐ray based imaging research areas. The results obtained for the six cases using four publicly available Monte Carlo software packages are included in tabular form. In addition to a full description of all simulation conditions and results, a discussion and comparison of results among the Monte Carlo packages and the lessons learned during the compilation of these results are included. This abridged version of the report includes only an introductory description of the six cases and a brief example of the results of one of the cases. This work provides an investigator the necessary information to benchmark his/her Monte Carlo simulation software against the reference cases included here before performing his/her own novel research. In addition, an investigator entering the field of Monte Carlo simulations can use these descriptions and results as a self‐teaching tool to ensure that he/she is able to perform a specific simulation correctly. Finally, educators can assign these cases as learning projects as part of course objectives or training programs.

Sechopoulos, Ioannis↗

SU-E-T-188: Film Dosimetry Verification of Monte Carlo Generated Electron Treatment Plans

Purpose: The purpose of this study was to compare dose distributions from film measurements to Monte Carlo generated electron treatment plans. Irradiation with electrons offers the advantages of dose uniformity in the target volume and of minimizing the dose to deeper healthy tissue. Using the Monte Carlo algorithm will improve dose accuracy in regions with heterogeneities and irregular surfaces. Methods: Dose distributions from GafChromic{sup ™} EBT3 films were compared to dose distributions from the Electron Monte Carlo algorithm in the Eclipse{sup ™} radiotherapy treatment planning system. These measurements were obtained for 6MeV, 9MeV and 12MeV electrons at two depths. All phantoms studied were imported into Eclipse by CT scan. A 1 cm thick solid water template with holes for bonelike and lung-like plugs was used. Different configurations were used with the different plugs inserted into the holes. Configurations with solid-water plugs stacked on top of one another were also used to create an irregular surface. Results: The dose distributions measured from the film agreed with those from the Electron Monte Carlo treatment plan. Accuracy of Electron Monte Carlo algorithm was also compared to that of Pencil Beam. Dose distributions from Monte Carlo had much higher pass rates than distributions from Pencil Beam when compared to the film. The pass rate for Monte Carlo was in the 80%–99% range, where the pass rate for Pencil Beam was as low as 10.76%. Conclusion: The dose distribution from Monte Carlo agreed with the measured dose from the film. When compared to the Pencil Beam algorithm, pass rates for Monte Carlo were much higher. Monte Carlo should be used over Pencil Beam for regions with heterogeneities and irregular surfaces.

60 APPLIED LIFE SCIENCES↗

Comparison of internal dose estimates obtained using organ-level, voxel S value, and Monte Carlo techniques

Purpose: The authors’ objective was to compare internal dose estimates obtained using the Organ Level Dose Assessment with Exponential Modeling (OLINDA/EXM) software, the voxel S value technique, and Monte Carlo simulation. Monte Carlo dose estimates were used as the reference standard to assess the impact of patient-specific anatomy on the final dose estimate. Methods: Six patients injected with{sup 99m}Tc-hydrazinonicotinamide-Tyr{sup 3}-octreotide were included in this study. A hybrid planar/SPECT imaging protocol was used to estimate {sup 99m}Tc time-integrated activity coefficients (TIACs) for kidneys, liver, spleen, and tumors. Additionally, TIACs were predicted for {sup 131}I, {sup 177}Lu, and {sup 90}Y assuming the same biological half-lives as the {sup 99m}Tc labeled tracer. The TIACs were used as input for OLINDA/EXM for organ-level dose calculation and voxel level dosimetry was performed using the voxel S value method and Monte Carlo simulation. Dose estimates for {sup 99m}Tc, {sup 131}I, {sup 177}Lu, and {sup 90}Y distributions were evaluated by comparing (i) organ-level S values corresponding to each method, (ii) total tumor and organ doses, (iii) differences in right and left kidney doses, and (iv) voxelized dose distributions calculated by Monte Carlo and the voxel S value technique. Results: The S values for all investigated radionuclides used by OLINDA/EXM and the corresponding patient-specific S values calculated by Monte Carlo agreed within 2.3% on average for self-irradiation, and differed by as much as 105% for cross-organ irradiation. Total organ doses calculated by OLINDA/EXM and the voxel S value technique agreed with Monte Carlo results within approximately ±7%. Differences between right and left kidney doses determined by Monte Carlo were as high as 73%. Comparison of the Monte Carlo and voxel S value dose distributions showed that each method produced similar dose volume histograms with a minimum dose covering 90% of the volume (D90) agreeing within ±3%, on average. Conclusions: Several aspects of OLINDA/EXM dose calculation were compared with patient-specific dose estimates obtained using Monte Carlo. Differences in patient anatomy led to large differences in cross-organ doses. However, total organ doses were still in good agreement since most of the deposited dose is due to self-irradiation. Comparison of voxelized doses calculated by Monte Carlo and the voxel S value technique showed that the 3D dose distributions produced by the respective methods are nearly identical.

60 APPLIED LIFE SCIENCES↗

SU-F-T-575: Verification of a Monte-Carlo Small Field SRS/SBRT Dose Calculation System

Purpose: To develop a methodology for validation of a Monte-Carlo dose calculation model for robotic small field SRS/SBRT deliveries. Methods: In a robotic treatment planning system, a Monte-Carlo model was iteratively optimized to match with beam data. A two-part analysis was developed to verify this model. 1) The Monte-Carlo model was validated in a simulated water phantom versus a Ray-Tracing calculation on a single beam collimator-by-collimator calculation. 2) The Monte-Carlo model was validated to be accurate in the most challenging situation, lung, by acquiring in-phantom measurements. A plan was created and delivered in a CIRS lung phantom with film insert. Separately, plans were delivered in an in-house created lung phantom with a PinPoint chamber insert within a lung simulating material. For medium to large collimator sizes, a single beam was delivered to the phantom. For small size collimators (10, 12.5, and 15mm), a robotically delivered plan was created to generate a uniform dose field of irradiation over a 2×2cm{sup 2} area. Results: Dose differences in simulated water between Ray-Tracing and Monte-Carlo were all within 1% at dmax and deeper. Maximum dose differences occurred prior to dmax but were all within 3%. Film measurements in a lung phantom show high correspondence of over 95% gamma at the 2%/2mm level for Monte-Carlo. Ion chamber measurements for collimator sizes of 12.5mm and above were within 3% of Monte-Carlo calculated values. Uniform irradiation involving the 10mm collimator resulted in a dose difference of ∼8% for both Monte-Carlo and Ray-Tracing indicating that there may be limitations with the dose calculation. Conclusion: We have developed a methodology to validate a Monte-Carlo model by verifying that it matches in water and, separately, that it corresponds well in lung simulating materials. The Monte-Carlo model and algorithm tested may have more limited accuracy for 10mm fields and smaller.

60 APPLIED LIFE SCIENCES↗

SU-F-T-619: Dose Evaluation of Specific Patient Plans Based On Monte Carlo Algorithm for a CyberKnife Stereotactic Radiosurgery System

Purpose: This study will use Monte Carlo to simulate the Cyberknife system, and intend to develop the third-party tool to evaluate the dose verification of specific patient plans in TPS. Methods: By simulating the treatment head using the BEAMnrc and DOSXYZnrc software, the comparison between the calculated and measured data will be done to determine the beam parameters. The dose distribution calculated in the Raytracing, Monte Carlo algorithms of TPS (Multiplan Ver4.0.2) and in-house Monte Carlo simulation method for 30 patient plans, which included 10 head, lung and liver cases in each, were analyzed. The γ analysis with the combined 3mm/3% criteria would be introduced to quantitatively evaluate the difference of the accuracy between three algorithms. Results: More than 90% of the global error points were less than 2% for the comparison of the PDD and OAR curves after determining the mean energy and FWHM.The relative ideal Monte Carlo beam model had been established. Based on the quantitative evaluation of dose accuracy for three algorithms, the results of γ analysis shows that the passing rates (84.88±9.67% for head,98.83±1.05% for liver,98.26±1.87% for lung) of PTV in 30 plans between Monte Carlo simulation and TPS Monte Carlo algorithms were good. And the passing rates (95.93±3.12%,99.84±0.33% in each) of PTV in head and liver plans between Monte Carlo simulation and TPS Ray-tracing algorithms were also good. But the difference of DVHs in lung plans between Monte Carlo simulation and Ray-tracing algorithms was obvious, and the passing rate (51.263±38.964%) of γ criteria was not good. It is feasible that Monte Carlo simulation was used for verifying the dose distribution of patient plans. Conclusion: Monte Carlo simulation algorithm developed in the CyberKnife system of this study can be used as a reference tool for the third-party tool, which plays an important role in dose verification of patient plans. This work was supported in part by the grant from Chinese Natural Science Foundation (Grant No. 11275105). Thanks for the support from Accuray Corp.

60 APPLIED LIFE SCIENCES↗

Monte Carlo Techniques for Nuclear Systems - Theory Lectures

These are lecture notes for a Monte Carlo class given at the University of New Mexico. The following topics are covered: course information; nuclear eng. review & MC; random numbers and sampling; computational geometry; collision physics; tallies and statistics; eigenvalue calculations I; eigenvalue calculations II; eigenvalue calculations III; variance reduction; parallel Monte Carlo; parameter studies; fission matrix and higher eigenmodes; doppler broadening; Monte Carlo depletion; HTGR modeling; coupled MC and T/H calculations; fission energy deposition. Solving particle transport problems with the Monte Carlo method is simple - just simulate the particle behavior. The devil is in the details, however. These lectures provide a balanced approach to the theory and practice of Monte Carlo simulation codes. The first lectures provide an overview of Monte Carlo simulation methods, covering the transport equation, random sampling, computational geometry, collision physics, and statistics. The next lectures focus on the state-of-the-art in Monte Carlo criticality simulations, covering the theory of eigenvalue calculations, convergence analysis, dominance ratio calculations, bias in Keff and tallies, bias in uncertainties, a case study of a realistic calculation, and Wielandt acceleration techniques. The remaining lectures cover advanced topics, including HTGR modeling and stochastic geometry, temperature dependence, fission energy deposition, depletion calculations, parallel calculations, and parameter studies. This portion of the class focuses on using MCNP to perform criticality calculations for reactor physics and criticality safety applications. It is an intermediate level class, intended for those with at least some familiarity with MCNP. Class examples provide hands-on experience at running the code, plotting both geometry and results, and understanding the code output. The class includes lectures & hands-on computer use for a variety of Monte Carlo calculations. Beginning MCNP users are encouraged to review LA-UR-09-00380, "Criticality Calculations with MCNP: A Primer (3nd Edition)" (available at http:// mcnp.lanl.gov under "Reference Collection") prior to the class. No Monte Carlo class can be complete without having students write their own simple Monte Carlo routines for basic random sampling, use of the random number generator, and simplified particle transport simulation.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Nuclear Data Uncertainty Quantification for Adjoint-Weighted Quantities via XGPT in Monte Carlo Codes

The Total Monte Carlo method (TMC) is an innovative approach to nuclear data uncertainty propagation. In TMC, a large number of independent ENDF files are randomly generated starting from nuclear models parameters and their uncertainties. These files are then processed with NJOY or similar codes to produce a set of formatted continuous-energy cross sections (i.e., ACE files), and independent Monte Carlo neutron transport simulations are run with these different ACE files as input. Finally the distributions of the considered response functions are obtained directly from the results of the Monte Carlo runs. Despite several improvements of the TMC method, this approach is often considered an inefficient way of propagating nuclear data uncertainties, as it requires several computationally expensive Monte Carlo runs. Covariance-based methods present important advantages for several applications, especially when different sources of uncertainties are to be investigated. For this reason, sensitivity analysis and uncertainty quantification capabilities are being implemented in several Monte Carlo neutron transport codes. On the other hand, the fact that no artificial multi-group covariance matrices are required and that higher moments of the responses distributions can be easily obtained, are definite advantages of the TMC method. In an effort to overcome the limitations of both approaches, a new method, called eXtended Generalized Perturbation Theory (XGPT) is under development. When performing nuclear data uncertainty propagation, such XGPT approach does not require generating discretized (i.e., multigroup) covariance matrices, and allows to estimate higher moments of the response probability distributions, via linear perturbation theory. One of the promising applications of the XGPT method involves its adoption in conjunction with the Proper Orthogonal Decomposition (POD) of nuclear data, to build a Reduced Order Model (ROM) for the approximation of thousands of independent TMC results in a single criticality source simulation. In the present work, this approach is employed for the first time to evaluate the impact of uncertainty in nuclear data on adjoint-weighted quantities. The method has been implemented in an extended version of the Serpent-2.1.24 Monte Carlo code and is demonstrated on two fast spectrum benchmarks, focusing on {sup 208}Pb nuclear data. A new method for propagating nuclear data uncertainties in criticality source Monte Carlo simulations has been developed. This eXtended Generalized Perturbation Theory (XGPT) approach allows to accurately reproduce reference Total Monte Carlo (TMC) uncertainty estimates in a single criticality source simulation. The present work aims to demonstrate that the proposed method can quantify nuclear data uncertainty based on adjoint-weighted quantities. The XGPT approach was applied to evaluate the impact of {sup 208}Pb nuclear data uncertainties on the effective prompt neutron lifetime (l{sub eff}) in two fast spectrum systems. The obtained XGPT results show a good agreement with the reference TMC results, obtained from independent Monte Carlo runs with {sup 208}Pb nuclear data from random TENDL-2013 evaluations. The XGPT method, implemented in a extended version of the Serpent Monte Carlo code and adopted along with the Proper Orthogonal Decomposition (POD) of the nuclear data, is able to provide an accurate reduced order model to approximate thousands TMC results within a single run, ensuring a large saving in computational requirements. The effect of {sup 208}Pb nuclear data uncertainties appears to be rather large. This is consistent with previous studies that adopted the same input random evaluations data from TENDL. (authors)

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Investigation of Portable Event-Based Monte Carlo Transport Using the NVIDIA Thrust Library

Power consumption considerations are driving future high performance computing platforms toward many-core computing architectures. Los Alamos National Laboratory's Trinity machine, available in 2016, will use both Intel Xeon Haswell processors and Intel Xeon Phi Knights Landing many integrated core (MIC) architecture coprocessors. Lawrence Livermore National Laboratory's Sierra machine, available in 2018, will use an IBM PowerPC architecture along with Nvidia graphics processing unit (GPU) architecture accelerators. These different advanced architectures make the computing landscape in upcoming years complex. Traditional approaches to Monte Carlo transport do not work efficiently on these new computing platforms. MIC architectures require vectorization to operate efficiently, and vectorization is difficult to achieve in Monte Carlo transport. GPU architectures require additional code to explicitly use the hardware, requiring significant code changes or hardware specific branches in the source code. A significant challenge for Monte Carlo transport projects is to simultaneously support within a single source code base efficient simulations for both the current generation of architectures and the different advanced computing architectures. In order to address these challenges, two important changes are typically required: a new algorithmic approach for solving Monte Carlo transport, and explicit use of hardware specific software. In this paper, we describe initial research investigations of an event-based Monte Carlo transport algorithm implemented using the Nvidia Thrust library on a GPU for a Monte Carlo test code. The event-based algorithm targets many-core architectures by increasing SIMD (single instruction multiple data) parallelism, while Thrust potentially provides portable performance by allowing one source code base to compile code targeted for both CPUs and GPUs. We described preliminary investigations of portable event-based Monte Carlo algorithms implemented using the Nvidia Thrust library in a research Monte Carlo test code. We found that an explicit CUDA implementation of an event-based Monte Carlo algorithm performed significantly more efficiently than a Thrust implementation on GPU platforms, most likely as a result of additional flexibility in access to different memory spaces on the GPU. Additionally, we showed that on GPU platforms and at large enough problem sizes the event-based implementations perform more efficiently than the serial history-based implementation running on the host CPU. While investigating this problem, we also discovered that the performance of the event-based algorithm is affected by what tallies are being used. A zonal scalar flux tally requires atomic operations that significantly impacted the performance of the code, in some cases producing slowdowns instead of speedups. We decided to remove the tally in order to focus on the effectiveness of the event-based algorithm. Future work will be required to research more effective ways of handling such tallies. Additionally, we would like to consider new ways for optimizing both the Thrust and CUDA versions, in order to see how much performance we have yet to achieve. The potential trade-off between portability and performance was demonstrated in this investigation. Thrust provides both CPU and GPU versions of the code in one code base, but it does so at a cost. We discovered approximately a factor of 2-5 performance difference between Thrust and CUDA on each GPU platform we tested. Thrust provides a tool to access the GPUs with less effort and specialization, but it does so by giving up fine grained control where extra performance can be found for this application. Future work will be required to determine whether this performance differential between Thrust and CUDA can be reduced for event-based Monte Carlo transport. (authors)

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Efficient Subset Simulation using Hamiltonian Neural Network enhanced Markov Chain Monte Carlo Methods

The Monte Carlo method delivers an unbiased estimate of the probability of failure. However, the variance of the estimate depends on the number of evaluated samples. This number must be very large for estimations of a low probability of failure. If the evaluation of each sample is computationally expensive, the crude Monte Carlo simulation strategy is impracticable. Therefore, subset simulations are used to reduce the required number of evaluations. Subset simulations require a Markov Chain Monte Carlo sampler, such as the random walk Metropolis-Hastings algorithm. The algorithm, however, struggles with sampling in low-probability regions, especially if they are narrow. As a consequence, advanced Markov Chain Monte Carlo simulations have been developed. In particular, the Hamiltonian Monte Carlo method explores the target distribution rapidly. Driven by the idea of Hamiltonian dynamics, this sampler provides a non-random walk through the target distribution. The incorporation of subset simulation and Hamiltonian Monte Carlo methods has shown promising results for reliability analysis. One downside of the Hamiltonian Monte Carlo method is that gradient evaluations are computationally expensive, especially when dealing with high-dimensional problems and evaluating long trajectories. We show that integrating Hamiltonian neural networks in Hamiltonian Monte Carlo simulations significantly speeds up the sampling task. Furthermore, the enhancement of adaptive trajectory length within the Hamiltonian Monte Carlo results in the efficient proposal of the following states. Based on this recent enhancement, we provide a fast sampling strategy for subset simulations using Hamiltonian neural networks to replace the evaluation of the gradient and significantly speed up the Hamiltonian Monte Carlo simulation.

97 MATHEMATICS AND COMPUTING↗

A Hybrid Monte Carlo-Deterministic Method for Global Binary Stochastic Medium Transport Problems

Global deep-penetration transport problems are difficult to solve using traditional Monte Carlo techniques. In these problems, the scalar flux distribution is desired at all points in the spatial domain (global nature), and the scalar flux typically drops by several orders of magnitude across the problem (deep-penetration nature). As a result, few particle histories may reach certain regions of the domain, producing a relatively large variance in tallies in those regions. Implicit capture (also known as survival biasing or absorption suppression) can be used to increase the efficiency of the Monte Carlo transport algorithm to some degree. A hybrid Monte Carlo-deterministic technique has previously been developed by Cooper and Larsen to reduce variance in global problems by distributing particles more evenly throughout the spatial domain. This hybrid method uses an approximate deterministic estimate of the forward scalar flux distribution to automatically generate weight windows for the Monte Carlo transport simulation, avoiding the necessity for the code user to specify the weight window parameters. In a binary stochastic medium, the material properties at a given spatial location are known only statistically. The most common approach to solving particle transport problems involving binary stochastic media is to use the atomic mix (AM) approximation in which the transport problem is solved using ensemble-averaged material properties. The most ubiquitous deterministic model developed specifically for solving binary stochastic media transport problems is the Levermore-Pomraning (L-P) model. Zimmerman and Adams proposed a Monte Carlo algorithm (Algorithm A) that solves the Levermore-Pomraning equations and another Monte Carlo algorithm (Algorithm B) that is more accurate as a result of improved local material realization modeling. Recent benchmark studies have shown that Algorithm B is often significantly more accurate than Algorithm A (and therefore the L-P model) for deep penetration problems such as examined in this paper. In this research, we investigate the application of a variant of the hybrid Monte Carlo-deterministic method proposed by Cooper and Larsen to global deep penetration problems involving binary stochastic media. To our knowledge, hybrid Monte Carlo-deterministic methods have not previously been applied to problems involving a stochastic medium. We investigate two approaches for computing the approximate deterministic estimate of the forward scalar flux distribution used to automatically generate the weight windows. The first approach uses the atomic mix approximation to the binary stochastic medium transport problem and a low-order discrete ordinates angular approximation. The second approach uses the Levermore-Pomraning model for the binary stochastic medium transport problem and a low-order discrete ordinates angular approximation. In both cases, we use Monte Carlo Algorithm B with weight windows automatically generated from the approximate forward scalar flux distribution to obtain the solution of the transport problem.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Monte Carlo calculation of monitor unit for electron arc therapy

Purpose: Monitor unit (MU) calculations for electron arc therapy were carried out using Monte Carlo simulations and verified by measurements. Variations in the dwell factor (DF), source-to-surface distance (SSD), and treatment arc angle ({alpha}) were studied. Moreover, the possibility of measuring the DF, which requires gantry rotation, using a solid water rectangular, instead of cylindrical, phantom was investigated. Methods: A phase space file based on the 9 MeV electron beam with rectangular cutout (physical size=2.6x21 cm{sup 2}) attached to the block tray holder of a Varian 21 EX linear accelerator (linac) was generated using the EGSnrc-based Monte Carlo code and verified by measurement. The relative output factor (ROF), SSD offset, and DF, needed in the MU calculation, were determined using measurements and Monte Carlo simulations. An ionization chamber, a radiographic film, a solid water rectangular phantom, and a cylindrical phantom made of polystyrene were used in dosimetry measurements. Results: Percentage deviations of ROF, SSD offset, and DF between measured and Monte Carlo results were 1.2%, 0.18%, and 1.5%, respectively. It was found that the DF decreased with an increase in {alpha}, and such a decrease in DF was more significant in the {alpha} range of 0 deg. - 60 deg. than 60 deg. - 120 deg. Moreover, for a fixed {alpha}, the DF increased with an increase in SSD. Comparing the DF determined using the rectangular and cylindrical phantom through measurements and Monte Carlo simulations, it was found that the DF determined by the rectangular phantom agreed well with that by the cylindrical one within {+-}1.2%. It shows that a simple setup of a solid water rectangular phantom was sufficient to replace the cylindrical phantom using our specific cutout to determine the DF associated with the electron arc. Conclusions: By verifying using dosimetry measurements, Monte Carlo simulations proved to be an alternative way to perform MU calculations effectively for electron arc therapy. Since Monte Carlo simulations can generate a precalculated database of ROF, SSD offset, and DF for the MU calculation, with a reduction in human effort and linac beam-on time, it is recommended that Monte Carlo simulations be partially or completely integrated into the commissioning of electron arc therapy.

61 RADIATION PROTECTION AND DOSIMETRY↗