Search NASA⌕ Search

SEARCH · Search NASA

Results for “stochastic sampling”

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 307 records · Page 17

Optimizing the implementation of the target motion sampling temperature treatment technique - How fast can it get?

This article discusses the optimization of the target motion sampling (TMS) temperature treatment method, previously implemented in the Monte Carlo reactor physics code Serpent 2. The TMS method was introduced in [1] and first practical results were presented at the PHYSOR 2012 conference [2]. The method is a stochastic method for taking the effect of thermal motion into account on-the-fly in a Monte Carlo neutron transport calculation. It is based on sampling the target velocities at collision sites and then utilizing the 0 K cross sections at target-at-rest frame for reaction sampling. The fact that the total cross section becomes a distributed quantity is handled using rejection sampling techniques. The original implementation of the TMS requires 2.0 times more CPU time in a PWR pin-cell case than a conventional Monte Carlo calculation relying on pre-broadened effective cross sections. In a HTGR case examined in this paper the overhead factor is as high as 3.6. By first changing from a multi-group to a continuous-energy implementation and then fine-tuning a parameter affecting the conservativity of the majorant cross section, it is possible to decrease the overhead factors to 1.4 and 2.3, respectively. Preliminary calculations are also made using a new and yet incomplete optimization method in which the temperature of the basis cross section is increased above 0 K. It seems that with the new approach it may be possible to decrease the factors even as low as 1.06 and 1.33, respectively, but its functionality has not yet been proven. Therefore, these performance measures should be considered preliminary. (authors)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Parallel replica dynamics method for bistable stochastic reaction networks: Simulation and sensitivity analysis

Stochastic reaction networks that exhibit bistable behavior are common in systems biology, materials science, and catalysis. Sampling of stationary distributions is crucial for understanding and characterizing the long-time dynamics of bistable stochastic dynamical systems. However, simulations are often hindered by the insufficient sampling of rare transitions between the two metastable regions. In this paper, we apply the parallel replica method for a continuous time Markov chain in order to improve sampling of the stationary distribution in bistable stochastic reaction networks. The proposed method uses parallel computing to accelerate the sampling of rare transitions. Furthermore, it can be combined with the path-space information bounds for parametric sensitivity analysis. With the proposed methodology, we study three bistable biological networks: the Schlögl model, the genetic switch network, and the enzymatic futile cycle network. We demonstrate the algorithmic speedup achieved in these numerical benchmarks. More significant acceleration is expected when multi-core or graphics processing unit computer architectures and programming tools such as CUDA are employed

97 MATHEMATICS AND COMPUTING↗

A stochastic diffusion process for Lochner's generalized Dirichlet distribution

The method of potential solutions of Fokker-Planck equations is used to develop a transport equation for the joint probability of N stochastic variables with Lochner’s generalized Dirichlet distribution as its asymptotic solution. Individual samples of a discrete ensemble, obtained from the system of stochastic differential equations, equivalent to the Fokker-Planck equation developed here, satisfy a unit-sum constraint at all times and ensure a bounded sample space, similarly to the process developed in for the Dirichlet distribution. Consequently, the generalized Dirichlet diffusion process may be used to represent realizations of a fluctuating ensemble of N variables subject to a conservation principle. Compared to the Dirichlet distribution and process, the additional parameters of the generalized Dirichlet distribution allow a more general class of physical processes to be modeled with a more general covariance matrix.

97 MATHEMATICS AND COMPUTING↗

Strategic Placement and Sizing of Distributed Generation for Resilience Enhancement of Distribution Grids With Microgrid Formation

The rise in frequency and severity of extreme weather events highlights the need for resilient power distribution networks. Microgrids can help improve the resilience of distribution grids by providing continuous power supply using local distribution generation (DG) when the distribution grid fails. In this paper, we propose an approach for optimal placement and sizing of DG to form multiple microgrids throughout the distribution network by restoration actions such as switching operations in case of distribution grid outages caused by extreme weather events. Considering the randomness of damaged distribution lines, the DG placement and sizing problem is formulated as a two-stage stochastic mixed-integer program, with the first stage determining the placement and size of DG, and the second stage focusing on minimizing the amount of load shedding through network restoration and microgrid formations for each scenario. Due to the large number of scenarios, the sample average approximation (SAA) method is employed to solve the problem. The results of case studies on a modified IEEE 33 bus distribution grid demonstrate the effectiveness of the proposed DG placement and sizing strategy in improving the resilience of distribution grids by allowing the formation of multiple microgrids. In addition, the robustness and accuracy of the SAA method are validated through various case studies.

Distributed generation planning↗

A two-stage stochastic optimization model for scheduling electric vehicle charging loads to relieve distribution-system constraints

Electric vehicles (EVs) hold promise to improve the energy efficiency and environmental impacts of transportation. However, widespread EV use can impose significant stress on electricity-distribution systems due to their added charging loads. This paper proposes a centralized EV charging-control model, which schedules the charging of EVs that have flexibility. This flexibility stems from EVs that are parked at the charging station for a longer duration of time than is needed to fully recharge the battery. The model is formulated as a two-stage stochastic optimization problem. The model captures the use of distributed energy resources and uncertainties around EV arrival times and charging demands upon arrival, non-EV loads on the distribution system, energy prices, and availability of energy from the distributed energy resources. We use a Monte Carlo-based sample-average approximation technique and an L-shaped method to solve the resulting optimization problem efficiently. We also apply a sequential sampling technique to dynamically determine the optimal size of the randomly sampled scenario tree to give a solution with a desired quality at minimal computational cost. Here, we demonstrate the use of our model on a Central-Ohio-based case study. We show the benefits of the model in reducing charging costs, negative impacts on the distribution system, and unserved EV-charging demand compared to simpler heuristics. Lastly, we also conduct sensitivity analyses, to show how the model performs and the resulting costs and load profiles when the design of the station or EV-usage parameters are changed.

33 ADVANCED PROPULSION SYSTEMS↗

Adaptive Sampling using Support Vector Machines

Reliability/safety analysis of stochastic dynamic systems (e.g., nuclear power plants, airplanes, chemical plants) is currently performed through a combination of Event-Tress and Fault-Trees. However, these conventional methods suffer from certain drawbacks: • Timing of events is not explicitly modeled • Ordering of events is preset by the analyst • The modeling of complex accident scenarios is driven by expert-judgment For these reasons, there is currently an increasing interest into the development of dynamic PRA methodologies since they can be used to address the deficiencies of conventional methods listed above.

99 GENERAL AND MISCELLANEOUS↗

Development and Integration of a Stochastic Clad Damage Propagation Model into PRONGHORN-SC Subchannel Analysis Code

The failure of fuel pins in nuclear reactors is intrinsically stochastic. Typically, a combination of variation in manufacturing that affects the material characteristics and the fuel assembly dimensions, variation in operating conditions, such as local power, coolant flow rate, and irradiation induced changes in material properties lead to a large uncertainty in failure margin of the fuel pins. Failure, therefore, may occur in exceptional pins with adverse combinations of these variations. Upon a metal fuel pin (U-Pu-Zr/HT9) failure, depressurization of the fuel pin takes place by release of fission gas, liquid sodium bond, and potentially solid fuel particles or molten/eutectic fuel droplets through the hole in cladding. The effect of a fission gas jet on neighbor fuel pins and possible propagation of a clad damage during normal operation was studied experimentally in 1970s and it was found that the post-failure fission gas jet insulates the jet impingement area of the target fuel pin surface and could increase the target pin’s surface temperature by as much as 100 – 200 K during the failed pin depressurization. It was concluded that the effect should not lead to fuel pin failure propagation during normal operation. In accident scenarios of sodium and lead fast reactors such as Unprotected Loss-Of-Flow (ULOF) or Unprotected Transient Over Power (UTOP), the fuel pins can be subjected to higher clad temperatures and fuel pin pressures or fuel clad mechanical/chemical interaction where thermal creep margin becomes significantly lower compared to the normal operation conditions. Therefore, possible stochastic failure and the post-failure fission gas/fuel jet impingement could be critical in order to predict fuel pin failure propagation. Pin depressurization due to fission gas release may degrade the heat transfer by formation of a gas blanket on a neighboring pin surface, which is a local phenomenon, and by causing coolant flow deceleration and starvation, which could affect a surrounding region as well. Furthermore, the potential presence of solid fuel particles or molten fuel at the time of clad failure could boost post-failure jet induced degradation even further. The present study models the U-Pu-Zr/HT9 metal fuel pin failure and stochastic clad damage propagation by biased sampling based on a Cumulative Damage Fraction (CDF) type clad failure criterion and the normal distribution of fuel failure probability density as a function of logarithm of Cumulative Damage Fraction. In addition, the effect of post-failure fission gas jet on heat transfer degradation is modeled for the target pins. This model is called stochastic Clad Damage Propagation (CDAP). The CDAP model is now fully integrated into developmental version of PRONGHORN-SC subchannel analysis code, allowing for modeling local failures and its propagation potential. Section 2 describes the components of the CDAP models. Section 3 describes the model implementation to PRONGHORN-SC and input specifications. Section 4 describes the CDAP model validation coupled to PRONGHORN-SC.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Analysis of the Space Propulsion System Problem Using RAVEN

This paper presents the solution of the space propulsion problem using a PRA code currently under development at Idaho National Laboratory (INL). RAVEN (Reactor Analysis and Virtual control ENviroment) is a multi-purpose Probabilistic Risk Assessment (PRA) software framework that allows dispatching different functionalities. It is designed to derive and actuate the control logic required to simulate the plant control system and operator actions (guided procedures) and to perform both Monte- Carlo sampling of random distributed events and Event Tree based analysis. In order to facilitate the input/output handling, a Graphical User Interface (GUI) and a post-processing data-mining module are available. RAVEN allows also to interface with several numerical codes such as RELAP5 and RELAP-7 and ad-hoc system simulators. For the space propulsion system problem, an ad-hoc simulator has been developed and written in python language and then interfaced to RAVEN. Such simulator fully models both deterministic (e.g., system dynamics and interactions between system components) and stochastic behaviors (i.e., failures of components/systems such as distribution lines and thrusters). Stochastic analysis is performed using random sampling based methodologies (i.e., Monte-Carlo). Such analysis is accomplished to determine both the reliability of the space propulsion system and to propagate the uncertainties associated to a specific set of parameters. As also indicated in the scope of the benchmark problem, the results generated by the stochastic analysis are used to generate risk-informed insights such as conditions under witch different strategy can be followed.

97 MATHEMATICS AND COMPUTING↗

Bayesian sparse learning with preconditioned stochastic gradient MCMC and its applications

Deep neural networks have been successfully employed in an extensive variety of research areas, including solving partial differential equations. Despite its significant success, there are some challenges in effectively training DNN, such as avoiding overfitting in over-parameterized DNNs and accelerating the optimization in DNNs with pathological curvature. Here, we propose a Bayesian type sparse deep learning algorithm. The algorithm utilizes a set of spike-and-slab priors for the parameters in the deep neural network. The hierarchical Bayesian mixture will be trained using an adaptive empirical method. That is, one will alternatively sample from the posterior using preconditioned stochastic gradient Langevin Dynamics (PSGLD), and optimize the latent variables via stochastic approximation. The sparsity of the network is achieved while optimizing the hyperparameters with adaptive searching and penalizing. A popular SG-MCMC approach is Stochastic gradient Langevin dynamics (SGLD). However, considering the complex geometry in the model parameter space in nonconvex learning, updating parameters using a universal step size in each component as in SGLD may cause slow mixing. To address this issue, we apply a computationally manageable preconditioner in the updating rule, which provides a step-size parameter to adapt to local geometric properties. Moreover, by smoothly optimizing the hyperparameter in the preconditioning matrix, our proposed algorithm ensures a decreasing bias, which is introduced by ignoring the correction term in the preconditioned SGLD. According to the existing theoretical framework, we show that the proposed algorithm can asymptotically converge to the correct distribution with a controllable bias under mild conditions. Numerical tests are performed on both synthetic regression problems and learning solutions of elliptic PDE, which demonstrate the accuracy and efficiency of the present work.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Improved Sampling Algorithms in the Risk-Informed Safety Margin Characterization Toolkit

The RISMC approach is developing advanced set of methodologies and algorithms in order to perform Probabilistic Risk Analyses (PRAs). In contrast to classical PRA methods, which are based on Event-Tree and Fault-Tree methods, the RISMC approach largely employs system simulator codes applied to stochastic analysis tools. The basic idea is to randomly perturb (by employing sampling algorithms) timing and sequencing of events and internal parameters of the system codes (i.e., uncertain parameters) in order to estimate stochastic parameters such as core damage probability. This approach applied to complex systems such as nuclear power plants requires to perform a series of computationally expensive simulation runs given a large set of uncertain parameters. These types of analysis are affected by two issues. Firstly, the space of the possible solutions (a.k.a., the issue space or the response surface) can be sampled only very sparsely, and this precludes the ability to fully analyze the impact of uncertainties on the system dynamics. Secondly, large amounts of data are generated and tools to generate knowledge from such data sets are not yet available. This report focuses on the first issue and in particular employs novel methods that optimize the information generated by the sampling process by sampling unexplored and risk-significant regions of the issue space: adaptive (smart) sampling algorithms. They infer system response from surrogate models constructed from existing samples and predict the most relevant location of the next sample. It is therefore possible to understand features of the issue space with a small number of carefully selected samples. In this report, we will present how it is possible to perform adaptive sampling using the RISMC toolkit and highlight the advantages compared to more classical sampling approaches such Monte-Carlo. We will employ RAVEN to perform such statistical analyses using both analytical cases but also another RISMC code: RELAP-7.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Stochastic Programming Approach for Resource Selection under Demand Uncertainty

Cost-efficient selection and scheduling of a subset of geographically distributed resources to meet the demands of a scientific workflow is a challenging problem. The problem is exacerbated by uncertainties in demand and availability of resources. In this paper, we present a stochastic optimization based framework for robust decision making in the selection of distributed resources over a planning horizon under demand uncertainty. We present a novel two-stage stochastic programming model for resource selection, and implement an L-shaped decomposition algorithm to solve this model. A Sample Average Approximation algorithm is integrated to enable stochastic optimization to solve problems with a large number of scenarios. Using the metric of stochastic solution, we demonstrate up to 30% cost reduction relative to solutions without explicit consideration of demand uncertainty for a 24-month problem. We also demonstrate up to 54% cost reduction relative to a previously developed solution for a 36-month problem. We further argue that the composition of resources selected is superior to solutions computed without explicit consideration of uncertainties. Given the importance of resource selection and scheduling of complex scientific workflows, especially in the context of commercial cloud computing, we believe that our novel stochastic programming framework will benefit many researchers as well as users of distributed computing resources.

scientific workflows, task scheduling, optimizatio↗

Numerical Uncertainty Estimation for Stochastic Particle-in-Cell Simulations Applied to Verification and Validation.

Systematic veri cation and validation (V&V) is necessary to establish the credibility for high consequence simulations. In this paper, we focus on a radiation-induced plasma experimental validation exercise for simulations which uses both numerical error estimation and input parameter uncertainty quanti cation to provide a direct comparison between Particle-In-Cell (PIC) plasma simulations and experiments. This approach demonstrates how careful validation can uncover missing physics in the simulation. Three di erent validation examples are shown; a vacuum space charge limited cavity, a gas lled space charge limited cavity, and a vacuum non space charge limited cavity. Two of the example are picked to show the importance of error estimation in uncovering inaccuracy/incomplete simulation models. We also report on a newly-developed numerical error estimation approach, StREEQ, which is a notable improvement to past approaches. In the StREEQ approach, a multi- tting scheme based on L 1 , L 2 , and L$\infty$ error norms and alternate weightings is used to propagate uncertainties in the relative importance of outliers and coarse/re ned discretization levels. Bootstrap sampling is used to represent the stochasticity in the response data. The resulting method appears to robustly and conservatively predict the fully-converged response within estimated numerical error bounds for stochastic simulations. The StREEQ approach is demonstrated on two related prototype electron diode problems, and preliminary results are reported for a radiation-induced plasma simulation.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

A generalized polynomial chaos based ensemble Kalman filter with high accuracy

As one of the most adopted sequential data assimilation methods in many areas, especially those involving complex nonlinear dynamics, the ensemble Kalman filter (EnKF) has been under extensive investigation regarding its properties and efficiency. Compared to other variants of the Kalman filter (KF), EnKF is straightforward to implement, as it employs random ensembles to represent solution states. This, however, introduces sampling errors that affect the accuracy of EnKF in a negative manner. Though sampling errors can be easily reduced by using a large number of samples, in practice this is undesirable as each ensemble member is a solution of the system of state equations and can be time consuming to compute for large-scale problems. In this paper we present an efficient EnKF implementation via generalized polynomial chaos (gPC) expansion. The key ingredients of the proposed approach involve (1) solving the system of stochastic state equations via the gPC methodology to gain efficiency; and (2) sampling the gPC approximation of the stochastic solution with an arbitrarily large number of samples, at virtually no additional computational cost, to drastically reduce the sampling errors. The resulting algorithm thus achieves a high accuracy at reduced computational cost, compared to the classical implementations of EnKF. Numerical examples are provided to verify the convergence property and accuracy improvement of the new algorithm. We also prove that for linear systems with Gaussian noise, the first-order gPC Kalman filter method is equivalent to the exact Kalman filter.

97 MATHEMATICS AND COMPUTING↗

Stochastic projective splitting

Here, we present a new, stochastic variant of the projective splitting (PS) family of algorithms for inclusion problems involving the sum of any finite number of maximal monotone operators. This new variant uses a stochastic oracle to evaluate one of the operators, which is assumed to be Lipschitz continuous, and (deterministic) resolvents to process the remaining operators. Our proposal is the first version of PS with such stochastic capabilities. We envision the primary application being machine learning (ML) problems, with the method’s stochastic features facilitating “mini-batch” sampling of datasets. Since it uses a monotone operator formulation, the method can handle not only Lipschitz-smooth loss minimization, but also min–max and noncooperative game formulations, with better convergence properties than the gradient descent-ascent methods commonly applied in such settings. The proposed method can handle any number of constraints and nonsmooth regularizers via projection and proximal operators. We prove almost-sure convergence of the iterates to a solution and a convergence rate result for the expected residual, and close with numerical experiments on a distributionally robust sparse logistic regression problem.

97 MATHEMATICS AND COMPUTING↗

Connectivity-informed drainage network generation using deep convolution generative adversarial networks

Abstract Stochastic network modeling is often limited by high computational costs to generate a large number of networks enough for meaningful statistical evaluation. In this study, Deep Convolutional Generative Adversarial Networks (DCGANs) were applied to quickly reproduce drainage networks from the already generated network samples without repetitive long modeling of the stochastic network model, Gibb’s model. In particular, we developed a novel connectivity-informed method that converts the drainage network images to the directional information of flow on each node of the drainage network, and then transforms it into multiple binary layers where the connectivity constraints between nodes in the drainage network are stored. DCGANs trained with three different types of training samples were compared; (1) original drainage network images, (2) their corresponding directional information only, and (3) the connectivity-informed directional information. A comparison of generated images demonstrated that the novel connectivity-informed method outperformed the other two methods by training DCGANs more effectively and better reproducing accurate drainage networks due to its compact representation of the network complexity and connectivity. This work highlights that DCGANs can be applicable for high contrast images common in earth and material sciences where the network, fractures, and other high contrast features are important.

42 ENGINEERING↗

Stabilized multilevel Monte Carlo method for stiff stochastic differential equations

A multilevel Monte Carlo (MLMC) method for mean square stable stochastic differential equations with multiple scales is proposed. For such problems, that we call stiff, the performance of MLMC methods based on classical explicit methods deteriorates because of the time step restriction to resolve the fastest scales that prevents to exploit all the levels of the MLMC approach. We show that by switching to explicit stabilized stochastic methods and balancing the stabilization procedure simultaneously with the hierarchical sampling strategy of MLMC methods, the computational cost for stiff systems is significantly reduced, while keeping the computational algorithm fully explicit and easy to implement. Numerical experiments on linear and nonlinear stochastic differential equations and on a stochastic partial differential equation illustrate the performance of the stabilized MLMC method and corroborate our theoretical findings.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Accurate Atomic Correlation and Total Energies for Correlation Consistent Effective Core Potentials

Very recently, we introduced a set of correlation consistent effective core potentials (ccECPs) constructed within full many-body approaches. By employing significantly more accurate correlated approaches, we were able to reach a new level of accuracy for the resulting effective core Hamiltonians. We also strived for simplicity of use and easy transferability into a variety of electronic structure methods in quantum chemistry and condensed matter physics. Here, as a reference for future use, we present exact or nearly exact total energy calculations for these ccECPs. The calculations cover H–Kr elements and are based on the state-of-the-art configuration interaction (CI), coupled-cluster (CC), and quantum Monte Carlo (QMC) calculations with systematically eliminated/improved errors. In particular, we carry out full CI/CCSD(T)/CCSDT(Q) calculations with cc-pVnZ with up to n = 6 basis sets and we estimate the complete basis set limits. Using combinations of these approaches, we achieved an accuracy of ≈1–10 mHa for K–Zn atoms and ≈0.1–0.3 mHa for all other elements—within about 1% or better of the ccECP total correlation energies. We also estimate the corresponding kinetic energies within the feasible limit of full CI calculations. In order to provide data for QMC calculations, we include fixed-node diffusion Monte Carlo energies for each element that give quantitative insights into the fixed-node biases for single-reference trial wave functions. The results offer a clear benchmark for future high-accuracy calculations in a broad variety of correlated wave function methods such as CI and CC as well is in stochastic approaches such as real space sampling QMC.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Efficient molecular dynamics using geodesic integration and solvent–solute splitting

We present an approach to Langevin dynamics in the presence of holonomic constraints based on decomposition of the system into components representing geodesic flow, constrained impulse and constrained diffusion. We show that a particular ordering of the components results in an integrator that is an order of magnitude more accurate for configurational averages than existing alternatives. Moreover, by combining the geodesic integration method with a solvent–solute force splitting, we demonstrate that stepsizes of at least 8 fs can be used for solvated biomolecules with high sampling accuracy and without substantially altering diffusion rates, approximately increasing by a factor of two the efficiency of molecular dynamics sampling for such systems. The methods described in this article are easily implemented using the standard apparatus of modern simulation codes.

97 MATHEMATICS AND COMPUTING↗