Tom Burr | Los Alamos National Laboratory (original) (raw)

Papers by Tom Burr

arXiv (Cornell University), Aug 2, 2009

The paper presents a multiplicative bias reduction estimator for nonparametric regression. The ap... more The paper presents a multiplicative bias reduction estimator for nonparametric regression. The approach consists to apply a multiplicative bias correction to an oversmooth pilot estimator. In Burr et al. [2010], this method has been tested to estimate energy spectra. For such data set, it was observed that the method allows to decrease bias with negligible increase in variance. In this paper, we study the asymptotic properties of the resulting estimate and prove that this estimate has zero asymptotic bias and the same asymptotic variance as the local linear estimate. Simulations show that our asymptotic results are available for modest sample sizes.

In this paper we develop and investigate several criteria for assessing how well a proposed spect... more In this paper we develop and investigate several criteria for assessing how well a proposed spectral form fits observed spectra. We consider the classical improved figure of merit (FOM) along with several modifications, as well as criteria motivated by Poisson regression from the statistical literature. We also develop a new FOM that is based on the statistical idea of the bootstrap. A spectral simulator has been developed to assess the performance of these different criteria under multiple data configurations.

Physics of Plasmas, 2021

A rotating tokamak plasma can interact resonantly with the external helical magnetic perturbation... more A rotating tokamak plasma can interact resonantly with the external helical magnetic perturbations, also known as error fields. This can lead to locking and then to disruptions. We leverage machine learning (ML) methods to predict the locking events. We use a coupled third-order nonlinear ordinary differential equation model to represent the interaction of the magnetic perturbation and the plasma rotation with the error field. This model is sufficient to describe qualitatively the locking and unlocking bifurcations. We explore using ML algorithms with the simulation data and experimental data, focusing on the methods that can be used with sparse datasets. These methods lead to the possibility of the avoidance of locking in real-time operations. We describe the operational space in terms of two control parameters: the magnitude of the error field and the rotation frequency associated with the momentum source that maintains the plasma rotation. The outcomes are quantified by order par...

Journal of the American Statistical Association, 2004

Currently deployed passive gamma and neutron detectors screen for illicit nuclear material. Archi... more Currently deployed passive gamma and neutron detectors screen for illicit nuclear material. Archived data can help evaluate special nuclear material detection probabilities and investigate several related issues, including (1) nuisance gamma alarms arising from naturally occurring radiation, (2) the impact of drifting neutron and gamma background rates, and (3) radioisotope identification performance. This paper illustrates roles for data mining to investigate issue (1) and briefly reviews data mining to investigate issues (2) and (3).

Journal of Physics: Conference Series, 2018

View the article online for updates and enhancements. You may also like Approximate Bayesian comp... more View the article online for updates and enhancements. You may also like Approximate Bayesian computation for forward modeling in cosmology Joël Akeret, Alexandre Refregier, Adam Amara et al.-Maximum likelihood versus likelihood-free quantum system identification in the atom maser

Nuclear Science and Engineering, 2015

An integrated nondestructive assay (NDA) system combining active (neutron generator) and passive ... more An integrated nondestructive assay (NDA) system combining active (neutron generator) and passive neutron detection and passive gamma (PG) detection is being analyzed in order to estimate the amount of plutonium, verify initial enrichment, burnup, and cooling time, and detect partial defects in a spent fuel assembly (SFA). Active signals are measured using the differential die-away (DDA), delayed neutron (DN), and delayed gamma (DG) techniques. Passive signals are measured using total neutron (TN) counts and both gross and spectral resolved gamma counts. To quantify how a system of several NDA techniques is expected to perform, all of the relevant NDA techniques listed above were simulated as a function of various reactor conditions such as initial enrichment, burnup, cooling time, assembly shuffling pattern, reactor operating conditions (including temperature, pressure, and the presence of burnable poisons) by simulating the NDA response for five sets of light water reactor assemblies. This paper compares the performance of several exploratory model-fitting options (including neural networks, adaptive regression with splines, iterative bias reduction smoothing, projection pursuit regression, and regression with quadratic terms and interaction terms) to relate data simulated with measurement and model error effects from various subsets of the NDA techniques to the total Pu mass. Isotope masses for SFAs and expected detector responses (DRs) for several NDA techniques are simulated using MCNP, and the DRs become inputs to the fitting process. Such responses include eight signals from DDA, one from DN, one from TN, and up to seven from PG; the DG signal will be examined separately. Results are summarized using the root-mean-squared estimation error for plutonium mass in held-out subsets of the data for a range of model and measurement error variances. Different simulation assumptions lead to different spent fuel libraries relating DRs to Pu mass. Some results for training with one library and testing with another library are also given. I. INTRODUCTION AND BACKGROUND The Next Generation Safeguards Initiative of the U.S. Department of Energy is sponsoring a multilaboratory/ university collaboration to quantify the plutonium (Pu) mass in, and to detect the diversion of pins from, spent nuclear fuel assemblies 1,2 (SFAs). Several techniques are being developed for this purpose, including recently studied neutron generator techniques that require a neutron generator on one side of the assembly to induce reactions within the assembly and produce particles that can be detected on the other sides. Such particles include active prompt neutrons using the differential die-away (DDA) technique, 3,4 active delayed neutrons (DNs) through the DN technique, 5,6 passive total neutron (TN)

Proceedings of 2002 Spring …, 2002

Remote detection and identification of chemicals in a scene is a challenging problem. We introduc... more Remote detection and identification of chemicals in a scene is a challenging problem. We introduce an approach that uses some of the image's pixels to establish the background characteristics while other pixels represent the target for which we seek to identify all ...

In order to recycle the nuclear resource and reduce the environmental burden, a closed fuel cycle... more In order to recycle the nuclear resource and reduce the environmental burden, a closed fuel cycle has been pursued in Japan. The total Pu throughput in a large reprocessing plant for mixed oxide (MOX) spent fuels produced from light water reactors becomes a tremendous quantity over time. Development of safeguards technologies and proliferation-resistant technologies is important to respond to nonproliferation concerns. Solution monitoring (SM) is currently used as an additional safeguards measure to confirm declared operations and to complement near-real-time accounting (NRTA) and containment and surveillance (C/S). Recent quantitative evaluations of SM have shown high detection probability (DP) for abrupt loss and moderate DP for protracted loss. In these studies, DP evaluation with multivariate statistical analysis was proposed as a quantified C/S. Moreover, a bias estimation and subtraction method was proposed to reduce the systematic error components for nuclear material account...

Applied Mathematics, 2021

Recent implementations of Bayesian approaches are one of the largest advances in phylogenetic tre... more Recent implementations of Bayesian approaches are one of the largest advances in phylogenetic tree estimation in the last 10 years. Markov chain Monte Carlo (MCMC) is used in these new approaches to estimate the Bayesian posterior probability for each tree topology of interest. Our goal is to assess the confidence in the estimated tree (particularly in whether prespecified groups are monophyletic) using MCMC and to compare the Bayesian estimate of confidence to a bootstrap-based estimate of confidence. We compare the Bayesian posterior probability to the bootstrap probability for specified groups in two real sets of influenza sequences and two sets of simulated sequences for our comparison. We conclude that the bootstrap estimate is adequate compared to the MCMC estimate except perhaps if the number of DNA sites is small.

Thomas L. Burr, CCS-6; Nicholas W. Hengartner, CCS-3; Steven C. Myers, N-2 The energy spectra of ... more Thomas L. Burr, CCS-6; Nicholas W. Hengartner, CCS-3; Steven C. Myers, N-2 The energy spectra of gamma-rays emitted by radioisotopes act as fingerprints that enable identification of the source. Such identification from low-resolution sodium iodide (NaI) detectors over short time periods is challenging for several reasons, including the Poisson fluctuations in the recorded counts. Smoothing the data over neighboring energy bins can reduce noise in the raw counts, at the cost of introducing a bias that de-emphasizes the peaks and valleys of the spectrum. This note describes a new two-stage smoothing procedure that uses a multiplicative bias correction for adjusting initial smoothed spectra. The benefits of this new method are illustrated on real spectra.

Nonparametric Statistics, 2018

Journal of Quality Technology, 2017

Inverse prediction is important in a variety of scientific and engineering applications, such as ... more Inverse prediction is important in a variety of scientific and engineering applications, such as to predict properties/characteristics of an object by using multiple measurements obtained from it. Inverse prediction can be accomplished by inverting parameterized forward models that relate the measurements (responses) to the properties/characteristics of interest. Sometimes forward models are computational/science based; but often, forward models are empirically based response surface models, obtained by using the results of controlled experimentation. For empirical models, it is important that the experiments provide a sound basis to develop accurate forward models in terms of the properties/characteristics (factors). While nature dictates the causal relationships between factors and responses, experimenters can control the complexity, accuracy, and precision of forward models constructed via selection of factors, factor levels, and the set of trials that are performed. Recognition of the uncertainty in the estimated forward models leads to an errors-in-variables approach for inverse prediction. The forward models (estimated by experiments or science based) can also be used to analyze how well candidate responses complement one another for inverse prediction over the range of the factor space of interest. One may find that some responses are complementary, redundant, or noninformative. Simple analysis and examples illustrate how an informative and discriminating subset of responses could be selected among candidates in cases where the number of responses that can be acquired during inverse prediction is limited by difficulty, expense, and/or availability of material.

Algorithms, 2015

Pattern recognition uses measurements from an input domain, X, to predict their labels from an ou... more Pattern recognition uses measurements from an input domain, X, to predict their labels from an output domain, Y. Image analysis is one setting where one might want to infer whether a pixel patch contains an object that is "manmade" (such as a building) or "natural" (such as a tree). Suppose the label for a pixel patch is "manmade"; if the label for a nearby pixel patch is then more likely to be "manmade" there is structure in the output domain that can be exploited to improve pattern recognition performance. Modeling P(X) is difficult because features between parts of the model are often correlated. Therefore, conditional random fields (CRFs) model structured data using the conditional distribution P(Y|X = x), without specifying a model for P(X), and are well suited for applications with dependent features. This paper has two parts. First, we overview CRFs and their application to pattern recognition in structured problems. Our primary examples are image analysis applications in which there is dependence among samples (pixel patches) in the output domain. Second, we identify research topics and present numerical examples.

Mathematical Biosciences and Engineering, 2009

Using daily counts of newly infected individuals, Wallinga and Teunis (WT) introduced a conceptua... more Using daily counts of newly infected individuals, Wallinga and Teunis (WT) introduced a conceptually simple method to estimate the number of secondary cases per primary case (Rt) for a given day. The method requires an estimate of the generation interval probability density function (pdf), which specifies the probabilities for the times between symptom onset in a primary case and symptom onset in a corresponding secondary case. Other methods to estimate Rt are based on explicit models such as the SIR model; therefore, one might expect the WT method to be more robust to departures from SIRtype behavior. This paper uses simulated data to compare the quality of daily Rt estimates based on a SIR model to those using the WT method for both structured (classical SIR assumptions are violated) and nonstructured (classical SIR assumptions hold) populations. By using detailed simulations that record the infection day of each new infection and the donor-recipient identities, the true Rt and the generation interval pdf is known with negligible error. We find that the generation interval pdf is time dependent in all cases, which agrees with recent results reported elsewhere. We also find that the WT method performs essentially the same in the structured populations (except for a spatial network) as it does in the nonstructured population. And, the WT method does as well or better than a SIR-model based method in three of the four structured populations. Therefore, even if the contact patterns are heterogeneous as in the structured populations evaluated here, the WT method provides reasonable estimates of Rt, as does the SIR method.