Hybrid biasing approaches for global variance reduction

Simulating nuclear well-logging devices with Monte Carlo methods is computationally challenging and requires significant variance reduction to compute detector responses with low statistical uncertainties in reasonable lengths of time. The consistent adjoint-driven importance sampling (CADIS) method, which provides consistent source and transport biasing parameters based on a deterministic adjoint (importance) function, has been demonstrated to be very effective for well-logging simulations and other deep-penetration problems. A recent extension to the CADIS method, FW-CADIS (forward-weighted CADIS), is designed to optimize the calculation of several tallies at once by using an adjoint function based on an adjoint source weighted by the inverse of the forward flux. These advanced variance reduction methods have been incorporated and automated into the MAVRIC sequence of SCALE, making them very easy to use. The CADIS and FW-CADIS methods are demonstrated and compared on simple benchmark models of both neutron- and photon-based well-logging devices. Both advanced variance reduction methods offer a substantial reduction in computing time, compared to analog simulation, for these applications.

The immense size and complex geometry of the ITER experimental fusion reactor require the development of special techniques that can accurately and efficiently perform neutronics simulations with minimal human effort. This paper shows the effect of the hybrid Monte Carlo (MC)/deterministic techniques—Consistent Adjoint Driven Importance Sampling (CADIS) and Forward-Weighted CADIS (FW-CADIS)—in enhancing the efficiency of the neutronics modeling of ITER and demonstrates the applicability of coupling these methods with computer-aided-design-based MC. Three quantities were calculated in this analysis: the total nuclear heating in the inboard leg of the toroidal field coils (TFCs), the prompt dose outside the biological shield, and the total neutron and gamma fluxes over a mesh tally covering the entire reactor. The use of FW-CADIS in estimating the nuclear heating in the inboard TFCs resulted in a factor of ˜275 increase in the MC figure of merit (FOM) compared with analog MC and a factor of ˜9 compared with the traditional methods of variance reduction. By providing a factor of ˜21 000 increase in the MC FOM, the radiation dose calculation showed how the CADIS method can be effectively used in the simulation of problems that are practically impossible using analog MC. The total flux calculation demonstrated the ability of FW-CADIS to simultaneously enhance the MC statistical precision throughout the entire ITER geometry. Collectively, these calculations demonstrate the ability of the hybrid techniques to accurately model very challenging shielding problems in reasonable execution times.

The application of the Automated Variance Reduction Generator (ADVANTG) code to accelerate MCNP neutron transport calculations in fusion-relevant geometries is presented. The ADVANTG code generates variance-reduction parameters using the Consistent Adjoint Driven Importance Sampling (CADIS) and Forward-Weighted Consistent Adjoint Driven Importance Sampling (FW-CADIS) methods based on deterministic transport calculations performed by the discrete ordinates code Denovo. The aim of ADVANTG is to reduce the MCNP computational time by automating the process of variance-reduction parameter generation. ADVANTG was tested on a simplified model of a JET-like tokamak that in spite of its simplicity retains all the major characteristics of such a tokamak. The performance of the nuclear data libraries provided with ADVANTG and of various other ADVANTG/Denovo settings on variance-reduction efficiency was tested. Several cases using deuterium-deuterium or deuterium-tritium (D-T) volumetric (plasma) sources and 252Cf or D-T point neutron sources were analyzed to find guidelines for successful use of the code for fusion applications. Additionally, the use of ADVANTG as a tool to identify major neutron pathways from the neutron source to the detector is demonstrated.

A new global variance reduction technique based on pseudo flux method

The hybrid Monte Carlo (MC)/deterministic techniques - Consistent Adjoint Driven Importance Sampling (CADIS) and Forward Weighted CADIS (FW-CADIS) - enable the full 3-D modeling of very large and complicated geometries. The ability of performing global MC calculations for nuclear parameters throughout the entire ITER reactor was demonstrated. The 2 m biological shield (bioshield) reduces the total prompt operational dose by six orders of magnitude. The divertor cryo-pump port results in a peaking factor of 120 in the prompt operational dose rate behind the bioshield of ITER. The equatorial port, plugged by 2 m of shielding, increases the prompt dose rate behind the bioshield by a factor of 47. The peak values of the prompt dose rates at the back surface of the bioshield were 240 μSv/hr and 94 μSv/hr corresponding to the regions behind the divertor cryo-pump port and the equatorial port, respectively.

Generalizable person re-identification aims to learn a model with only several labeled source domains that can perform well on unseen domains. Without access to the unseen domain, the feature statistics of the batch normalization (BN) layer learned from a limited number of source domains is doubtlessly biased for unseen domain. This would mislead the feature representation learning for unseen domain and deteriorate the generalizaiton ability of the model. In this paper, we propose a novel Debiased Batch Normalization via Gaussian Process approach (GDNorm) for generalizable person re-identification, which models the feature statistic estimation from BN layers as a dynamically self-refining Gaussian process to alleviate the bias to unseen domain for improving the generalization. Specifically, we establish a lightweight model with multiple set of domain-specific BN layers to capture the discriminability of individual source domain, and learn the corresponding parameters of the domain-specific BN layers. These parameters of different source domains are employed to deduce a Gaussian process. We randomly sample several paths from this Gaussian process served as the BN estimations of potential new domains outside of existing source domains, which can further optimize these learned parameters from source domains, and estimate more accurate Gaussian process by them in return, tending to real data distribution. Even without a large number of source domains, GDNorm can still provide debiased BN estimation by using the mean path of the Gaussian process, while maintaining low computational cost during testing. Extensive experiments demonstrate that our GDNorm effectively improves the generalization ability of the model on unseen domain.

The solid fuel thorium molten salt reactor (TMSR-SF1) is a 10-MWth fluoride-cooled pebble bed reactor. As a new reactor concept, one of the major limiting factors to reactor lifetime is radiation-induced material damage. The fast neutron flux (E > 0.1 MeV) can be used to assess possible radiation damage. Hence, a method for calculating high-resolution fast neutron flux distribution of the full-scale TMSR-SF1 reactor is required. In this study, a two-step subsection approach based on MCNP5 involving a global variance reduction method, referred to as forward-weighted consistent adjoint-driven importance sampling, was implemented to provide fast neutron flux distribution throughout the TMSR-SF1 facility. In addition, instead of using the general source specification cards, the user-provided SOURCE subroutine in MCNP5 source code was employed to implement a source biasing technique specialized for TMSR-SF1. In contrast to the one-step analog approach, the two-step subsection approach eliminates zero-scored mesh tally cells and obtains tally results with extremely uniform and low relative uncertainties. Furthermore, the maximum fast neutron fluxes of the main components in TMSR-SF1 are provided, which can be used for radiation damage assessment of the structural materials.

Global variance reduction method based on multi-group Monte Carlo adjoint calculation

A hybrid Gaussian process approach to robust economic model predictive control

This study compared hierarchical Bayesian, mixed-effects Gaussian process regression, and random forest models for predicting height to crown base (HCB) in Qinghai spruce (Picea crassifolia Kom.) forests using LiDAR-derived data. Both modeling approaches were applied to a dataset of 510 trees from 16 plots in northern China. The models incorporated tree-level variables (height, diameter at breast height, crown projection area) and plot-level spatial competition indices. Model performance was evaluated using leave-one-plot-out cross-validation. The Gaussian mixed-effects process model (with an RMSE of 1.59 and MAE of 1.25) slightly outperformed the hierarchical Bayesian model and the random forest model. Both models identified LiDAR-derived tree height, DBH, and LiDAR-derived crown projection area as primary factors influencing HCB. The spatial competition index (SCI) emerged as the most effective random effect, with the lowest AIC and BIC values, highlighting the importance of local competition dynamics in HCB formation. Uncertainty analysis revealed consistent patterns across the predicted values, with an average relative uncertainty of 33.89% for the Gaussian process model. These findings provide valuable insights for forest management and suggest that incorporating spatial competition indices can enhance HCB predictions.

Gaussian Process and Monte Carlo Approach for Occupant Behavior Modeling in Energy Simulation of Residential Buildings

A novel approach for the generation of complex humanoid walking sequences based on a combination of optimal control and learning of movement primitives

Different stellar phenomena affect radial velocities (RVs), causing variations large enough to make it difficult to identify planet signals from the stellar variability. RV variations caused by stellar oscillations and granulation can be reduced through some methods, but the impact of rotationally modulated magnetic activity on RV, due to stellar active regions, is harder to correct. New instrumentation promises an improvement in precision of one order of magnitude, from about 1 m s−1 to about 10 cm s−1. In this context, we report our first results from 24 spectroscopic Echelle SPectrograph for Rocky Exoplanets and Stable Spectroscopic Observations (ESPRESSO)/Very-Large Telescope observations of the solar twin star HIP 11915, spread over 60 nights. We used a Gaussian Process approach and found for HIP 11915 a RV residual rms scatter of about 20 cm s−1, representing an upper limit for the performance of ESPRESSO.

Analysis of inconsistent source sampling in monte carlo weight-window variance reduction methods

BackgroundEstimating the health effects of multi-pollutant mixtures is of increasing interest in environmental epidemiology. Recently, a new approach for estimating the health effects of mixtures, Bayesian kernel machine regression (BKMR), has been developed. This method estimates the multivariable exposure-response function in a flexible and parsimonious way, conducts variable selection on the (potentially high-dimensional) vector of exposures, and allows for a grouped variable selection approach that can accommodate highly correlated exposures. However, the application of this novel method has been limited by a lack of available software, the need to derive interpretable output in a computationally efficient manner, and the inability to apply the method to non-continuous outcome variables.MethodsThis paper addresses these limitations by (i) introducing an open-source software package in the R programming language, the bkmr R package, (ii) demonstrating methods for visualizing high-dimensional exposure-response functions, and for estimating scientifically relevant summaries, (iii) illustrating a probit regression implementation of BKMR for binary outcomes, and (iv) describing a fast version of BKMR that utilizes a Gaussian predictive process approach. All of the methods are illustrated using fully reproducible examples with the provided R code.ResultsApplying the methods to a continuous outcome example illustrated the ability of the BKMR implementation to estimate the health effects of multi-pollutant mixtures in the context of a highly nonlinear, biologically-based dose-response function, and to estimate overall, single-exposure, and interactive health effects. The Gaussian predictive process method led to a substantial reduction in the runtime, without a major decrease in accuracy. In the setting of a larger number of exposures and a dichotomous outcome, the probit BKMR implementation was able to correctly identify the variables included in the exposure-response function and yielded interpretable quantities on the scale of a latent continuous outcome or on the scale of the outcome probability.ConclusionsThis newly developed software, integrated suite of tools, and extended methodology makes BKMR accessible for use across a broad range of epidemiological applications in which multiple risk factors have complex effects on health.