Search

Neural networks have become very popular in surrogate modeling because of their ability to characterize arbitrary, high dimensional functions in a data driven fashion. This paper advocates for the training of surrogates that are consistent with the physical manifold -- i.e., predictions are always physically meaningful, and are cyclically consistent -- i.e., when the predictions of the surrogate, when passed through an independently trained inverse model give back the original input parameters. We find that these two consistencies lead to surrogates that are superior in terms of predictive performance, more resilient to sampling artifacts, and tend to be more data efficient. Using Inertial Confinement Fusion (ICF) as a test bed problem, we model a 1D semi-analytic numerical simulator and demonstrate the effectiveness of our approach. Code and data are available at https://github.com/rushilanirudh/macc/, 10 pages, 6 figures

Inertial confinement fusion (ICF) experiments are designed using computer simulations that are approximations of reality and therefore must be calibrated to accurately predict experimental observations. In this article, we propose a novel technique for calibrating from simulations to experiments, or from low fidelity simulations to high fidelity simulations, via “transfer learning” (TL). TL is a commonly used technique in the machine learning community, in which models trained on one task are partially retrained to solve a separate, but related task, for which there is a limited quantity of data. We introduce the idea of hierarchical TL, in which neural networks trained on low fidelity models are calibrated to high fidelity models, then to experimental data. This technique essentially bootstraps the calibration process, enabling the creation of models which predict high fidelity simulations or experiments with minimal computational cost. We apply this technique to a database of ICF simulations and experiments carried out at the Omega laser facility. TL with deep neural networks enables the creation of models that are more predictive of Omega experiments than simulations alone. The calibrated models accurately predict future Omega experiments, and are used to search for new, optimal implosion designs.

With the rapid adoption of machine learning techniques for large-scale applications in science and engineering comes the convergence of two grand challenges in visualization. First, the utilization of black box models (e.g., deep neural networks) calls for advanced techniques in exploring and interpreting model behaviors. Second, the rapid growth in computing has produced enormous datasets that require techniques that can handle millions or more samples. Although some solutions to these interpretability challenges have been proposed, they typically do not scale beyond thousands of samples, nor do they provide the high-level intuition scientists are looking for. Here, we present the first scalable solution to explore and analyze high-dimensional functions often encountered in the scientific data analysis pipeline. By combining a new streaming neighborhood graph construction, the corresponding topology computation, and a novel data aggregation scheme, namely topology aware datacubes , we enable interactive exploration of both the topological and the geometric aspect of high-dimensional data. Following two use cases from high-energy-density (HED) physics and computational biology, we demonstrate how these capabilities have led to crucial new insights in both applications.

We adopt a technique, known in the machine learning community as transfer learning, to reduce the bias of computer simulation using very sparse experimental data. Unlike the Bayesian calibration, which is commonly used to estimate the simulation bias, the transfer learning approach discussed in this article involves calculating an artificial neural network surrogate model of the simulations. Assuming that the simulation code correctly predicts the trends in the experimental data but it is subject to unknown biases, we then partially retrain, or transfer learn, the initial surrogate model to match the experimental data. This process eliminates the bias while still taking advantage of the physics relations learned from the simulation. Transfer learning can be easily adapted to a wide range of problems in science and engineering. In this article, we carry out numerical tests to investigate the applicability of this technique to predict the observable outcomes of inertial confinement fusion (ICF) experiments under new conditions. Using our synthetic validation data set, we demonstrate that an accurate predictive model can be built by retraining an initial surrogate model with experimental data volumes so small that they are relevant to the ICF problem. This opens up new opportunities for knowledge transfer and building predictive models in physics. After implementing transfer learning in a standard neural network, we successfully extended the method to a more complex, generative adversarial network architecture, which will be needed for predicting not only scalars but also diagnostic images in our future work.

Inertial confinement fusion implosions designed to have minimal fluid motion at peak compression often show significant linear flows in the laboratory, attributable per simulations to percent-level imbalances in the laser drive illumination symmetry. We present experimental results which intentionally varied the mode 1 drive imbalance by up to 4% to test hydrodynamic predictions of flows and the resultant imploded core asymmetries and performance, as measured by a combination of DT neutron spectroscopy and high-resolution x-ray core imaging. Neutron yields decrease by up to 50%, and anisotropic neutron Doppler broadening increases by 20%, in agreement with simulations. Furthermore, a tracer jet from the capsule fill-tube perturbation that is entrained by the hot-spot flow confirms the average flow speeds deduced from neutron spectroscopy.

As high-intensity short-pulse lasers that can operate at high-repetition-rate (HRR) (>10 Hz) come online around the world, the high-energy-density (HED) science they enable will experience a radical paradigm shift. The >10^3x increase in shot rate over today’s shot-per-hour drivers translates into dramatically faster data acquisition and more experiments, and thus the potential to significantly accelerate the advancement of HED science. We will present the vision and ongoing work to realize a HRR framework that allows for rapidly delivered optimal experiments by bringing together feedback laser control loops, high-throughput targetry and diagnostics, cognitive simulation, enhanced HED codes, and advanced data analytics.

The design space for inertial confinement fusion (ICF) experiments is vast and experiments are extremely expensive. Researchers rely heavily on computer simulations to explore the design space in search of high-performing implosions. However, ICF multiphysics codes must make simplifying assumptions, and thus deviate from experimental measurements for complex implosions. For more effective design and investigation, simulations require input from past experimental data to better predict future performance. In this work, we describe a cognitive simulation method for combining simulation and experimental data into a common, predictive model. This method leverages a machine learning technique called transfer learning, the process of taking a model trained to solve one task, and partially retraining it on a sparse dataset to solve a different, but related task. In the context of ICF design, neural network models trained on large simulation databases and partially retrained on experimental data, producing models that are far more accurate than simulations alone. We demonstrate improved model performance for a range of ICF experiments at the National Ignition Facility, and predict the outcome of recent experiments with less than ten percent error for several key observables. We discuss how the methods might be used to carry out a data-driven experimental campaign to optimize performance, illustrating the key product -- models that become increasingly accurate as data is acquired.

Predictive models that accurately emulate complex scientific processes can achieve speed-ups over numerical simulators or experiments and at the same time provide surrogates for improving the subsequent analysis. Consequently, there is a recent surge in utilizing modern machine learning methods to build data-driven emulators. In this work, we study an often overlooked, yet important, problem of choosing loss functions while designing such emulators. Popular choices such as the mean squared error or the mean absolute error are based on a symmetric noise assumption and can be unsuitable for heterogeneous data or asymmetric noise distributions. We propose Learn-by-Calibrating, a novel deep learning approach based on interval calibration for designing emulators that can effectively recover the inherent noise structure without any explicit priors. Using a large suite of use-cases, we demonstrate the efficacy of our approach in providing high-quality emulators, when compared to widely-adopted loss function choices, even in small-data regimes., The success of machine learning for scientific discovery normally depends on how well the inherent assumptions match the problem in hand. Here, Thiagarajan et al. alleviate this constraint by allowing the change of optimization criterion in a data-driven approach to emulate complex scientific processes.

Experiments and analysis in the 2 years since the 2015 Target Fabrication Meeting have resulted in further evolution of the requirements for high-performance layered implosions. This paper is a status update on the experimental program and supporting modeling, with emphasis on the implications for fabrication requirements. Previous work on the capsule support has continued, with various other support options being explored in experiments and modeling. Work also continues on ablator composition nonuniformities, with important new results from CH experiments on Omega, and the first three-dimensional X-ray transmission measurements of Be capsules on the National Ignition Facility. Work on hohlraums continues to include near-vacuum hohlraums and U hohlraums without a gold lining. Overall, the understanding that has been achieved, along with the progress in fabrication technology, represents good continuing progress toward the goal of fusion in the laboratory.

Nuclear reactions taking place in stars are not straightforward to study in laboratories on Earth. Now, inertial-confinement fusion implosion experiments are reported that mimic the conditions for the hydrogen-burning phase in main-sequence stars.

The achievement of inertial confinement fusion ignition on the National Ignition Facility relies on the collection and interpretation of a limited (and expensive) set of experimental data. These data are therefore supplemented with state-of-the-art multidimensional radiation-hydrodynamic simulations to provide a better understanding of implosion dynamics and behavior. We present a relatively large number (∼ 4000) of systematically perturbed 2D simulations to probe our understanding of low-mode fuel and ablator asymmetries seeded by asymmetric illumination. We find that Gaussian process surrogate models are able to predict both the total neutron yield and the degradation in performance due to asymmetries. The surrogates are then applied to simulations containing new sources of degradation to quantify the impact of the new source.

We present an introduction to deep learning for the high-brightness source community, covering methods and examples relevant to computational and experimental investigation. We focus on helping investigators select tools to accelerate entry into deep-learning-augmented research.

Training deep neural networks on large scientific data is a challenging task that requires enormous compute power, especially if no pre-trained models exist to initialize the process. We present a novel tournament method to train traditional as well as generative adversarial networks built on LBANN, a scalable deep learning framework optimized for HPC systems. LBANN combines multiple levels of parallelism and exploits some of the worlds largest supercomputers. We demonstrate our framework by creating a complex predictive model based on multi-variate data from high-energy-density physics containing hundreds of millions of images and hundreds of millions of scalar values derived from tens of millions of simulations of inertial confinement fusion. Our approach combines an HPC workflow and extends LBANN with optimized data ingestion and the new tournament-style training algorithm to produce a scalable neural network architecture using a CORAL-class supercomputer. Experimental results show that 64 trainers (1024 GPUs) achieve a speedup of 70.2 over a single trainer (16 GPUs) baseline, and an effective 109% parallel efficiency.

This paper presents a study on hotspot parameters in indirect-drive, inertially confined fusion implosions as they proceed through the self-heating regime. The implosions with increasing nuclear yield reach the burning-plasma regime, hotspot ignition, and finally propagating burn and ignition. These implosions span a wide range of alpha heating from a yield amplification of 1.7-2.5. We show that the hotspot parameters are explicitly dependent on both yield and velocity and that by fitting to both of these quantities the hotspot parameters can be fit with a single power law in velocity. The yield scaling also enables the hotspot parameters extrapolation to higher yields. This is important as various degradation mechanisms can occur on a given implosion at fixed implosion velocity which can have a large impact on both yield and the hotspot parameters. The yield scaling also enables the experimental dependence of the hotspot parameters on yield amplification to be determined. The implosions reported have resulted in the highest yield (1.73×10^{16}±2.6%), yield amplification, pressure, and implosion velocity yet reported at the National Ignition Facility.

To achieve hotspot ignition, an inertial confinement fusion implosion must achieve high hotspot pressure that is inertially confined by a dense shell of DT fuel. This requires a symmetric implosion having high in-flight shell velocity and high areal density at stagnation. The size of the driver and scale of the capsule required can be minimized by maintaining a high efficiency of energy coupling from the imploding shell to the hotspot. Significant 3D low mode asymmetries, however, are commonly observed in indirect-drive implosions and reduce the coupling of shell kinetic energy to the hotspot. To better quantify the magnitudes and impacts of shell density asymmetries, we have developed new analysis techniques and analytic models [Hurricane et al., Phys. Plasmas 27(6), 062704 (2020)]. To build confidence in the underlying data, we have also developed an analytic neutron transport model to cross-compare two independent measurements of asymmetry, which shows excellent agreement across shots for mode-1 (l = 1). This work also demonstrates that asymmetry can introduce potential sampling bias into down-scattered ratio measurements causing the solid-angle-average and uncertainty-weighted-average down-scattered ratios to differ significantly. Diagnosing asymmetries beyond mode-1 (l > 1) presents significant challenges. Using new diagnostic instruments and analysis techniques, however, evidence of significant Legendre mode P2 (l = 2, m = 0) and additional 3D asymmetries (l > 1, m ≠ 0) are beginning to emerge from the high precision activation diagnostic data (real-time nuclear activation detectors) and down-scattered neutron imaging data.

Experiments and analysis in the 3 years since the 2012 Target Fabrication Meeting have resulted in significant improvement in understanding of the requirements for high-performance layered implosio...

Inertial confinement fusion, based on laser-heating a deuterium–tritium mixture, is one of the approaches towards energy production from fusion reactions. Now, record energy-yield experiments are reported—bringing us closer to ignition conditions.

The shaping of the drive pulse in time is a key tool in the design of fusion experiments that use inertia to confine burning plasmas. It is directly related to the adiabat and compressibility of the DT fuel and the characteristics of the laser and target that are needed to ignite. With this in mind, we have performed experiments at the National Ignition Facility that test small changes in the shape of the pulse. In contrast to theory, we find implosions at lower adiabats can have reduced yield and areal density. We discuss implications to performance and the mechanism(s) that could be responsible.

Plasma jets, such as γ-ray burst jets, Herbig–Haro jets, μ-quasar jets, and active galactic nuclei jets, are found throughout the universe [S. Mendoza et al., Rev. Mex. Astron. Astrofis. 41, 453 (2005)]. Plasma jets are also present in indirect drive inertial confinement fusion experiments originating from the capsule's fill tube and occasionally from divots and voids in the capsules, particles on the exterior of the capsule, or from the tent holding the capsule in the target. This paper looks at two different gas-filled capsule implosions containing a plasma jet resulting from a capsule fill tube and fill channel, both of which utilized high density carbon ablators. Two models were developed, a drag and a snowplow model, which use the time-dependent motion of the injected mass through the hotspot to estimate the mass injected into the hotspot from the fill tube and channel, arriving at an average injected mass of ∼84.5 ± 25.5 ng for the first experiment and 91 ± 20 ng for the second experiment. Unlike previous methods to estimate fill tube injected mass, these techniques do not assume that the mixed mass is in thermal equilibrium with the hotspot or that the x-ray emission is only coming from within the hotspot itself. This paper also discusses the features seen in these experiments which include limb brightening in the shell for undoped ablators and flattening in the ablator from shadowing by the fill tube.

This paper analyzes x-ray-driven implosions that are designed to be less sensitive to 2D and 3D effects in Hohlraum and capsule physics. Key performance metrics including the burn-averaged ion temperature, hot-spot areal density, and fusion yield are found to agree with simulations where the design adiabat (internal pressure) is multiplied by a factor of 1.4. These results motivate the development of a simple model for interpreting experimental data, which is then used to quantify how improvements in compression could help achieve ignition.

Progress in inertial confinement fusion depends on the accurate interpretation of experiments that are complex and difficult to explain with simulations. Results could depend on small changes in the laser pulse or target or physics that are not fully understood or characterized. In this paper, we discuss an x-ray-driven platform [Baker et al., Phys. Rev. Lett. 121, 135001 (2018)] with fewer sources of degradation and find the fusion yield can be described as a physically motivated function of laser energy, target scale, and implosion symmetry. This platform and analysis could enable a more experimental approach to the study and optimization of implosion physics.

For many years, low mode asymmetry in inertially confined fusion (ICF) implosions has been recognized as a potential performance limiting factor, but analysis has been limited to using simulations and searching for data correlations. Herein, an analytically solvable model based upon the simple picture of an asymmetric piston is presented. Asymmetry of the shell driving the implosion, as opposed to asymmetry in the hot-spot, is key to the model. The model provides a unifying framework for the action of mode-1 shell asymmetry and the resulting connections between various diagnostic signatures. A key variable in the model is the shell asymmetry fraction, f, which is related to the areal density variation of the shell surrounding the hot-spot. It is shown that f is simply related to the observed hot-spot mode-1 velocity and to the concept of residual energy in an implosion. The model presented in this paper yields explicit expressions for the hot-spot diameter, stagnation pressure, hot-spot energy, inertial confinement-time, Lawson parameter, hot-spot temperature, and fusion yield under the action of mode-1 asymmetry. Agreement is found between the theory scalings when compared to ICF implosion data from the National Ignition Facility and to large ensembles of detailed simulations, making the theory a useful tool for interpreting data. The theory provides a basis for setting tolerable limits on asymmetry.

We describe the overall performance of the major indirect-drive inertial confinement fusion campaigns executed at the National Ignition Facility. With respect to the proximity to ignition, we can describe the performance of current experiments both in terms of no-burn ignition metrics (metrics based on the hydrodynamic performance of targets in the absence of alpha-particle heating) and in terms of the thermodynamic properties of the hotspot and dense fuel at stagnation—in particular, the hotspot pressure, temperature, and areal density. We describe a simple 1D isobaric model to derive these quantities from experimental observables and examine where current experiments lie with respect to the conditions required for ignition.

Computer simulations of high energy density science experiments are computationally challenging, consisting of multiple physics calculations including radiation transport, hydrodynamics, atomic physics, nuclear reactions, laser–plasma interactions, and more. To simulate inertial confinement fusion (ICF) experiments at high fidelity, each of these physics calculations should be as detailed as possible. However, this quickly becomes too computationally expensive even for modern supercomputers, and thus many simplifying assumptions are made to reduce the required computational time. Much of the research has focused on acceleration techniques for the various packages in multiphysics codes. In this work, we explore a novel method for accelerating physics packages via machine learning. The non-local thermodynamic equilibrium (NLTE) package is one of the most expensive calculations in the simulations of indirect drive inertial confinement fusion, taking several tens of percent of the total wall clock time. We explore the use of machine learning to accelerate this package, by essentially replacing the physics calculation with a deep neural network that has been trained to emulate the physics code. We demonstrate the feasibility of this approach on a simple problem and perform a side-by-side comparison of the physics calculation and the neural network inline in an ICF Hohlraum simulation. We show that the neural network achieves a 10× speed up in NLTE computational time while achieving good agreement with the physics code for several quantities of interest.

With the growing complexity of computational and experimental facilities, many scientific researchers are turning to machine learning (ML) techniques to analyze large scale ensemble data. With complexities such as multi-component workflows, heterogeneous machine architectures, parallel file systems, and batch scheduling, care must be taken to facilitate this analysis in a high performance computing (HPC) environment. In this paper, we present Merlin, a workflow framework to enable large ML-friendly ensembles of scientific HPC simulations. By augmenting traditional HPC with distributed compute technologies, Merlin aims to lower the barrier for scientific subject matter experts to incorporate ML into their analysis. In addition to its design, we describe some example applications that Merlin has enabled on leadership-class HPC resources, such as the ML-augmented optimization of nuclear fusion experiments and the calibration of infectious disease models to study the progression of and possible mitigation strategies for COVID-19., Comment: 28 pages, 9 figures; Submitted to FGCS

An important diagnostic value of a shot at the National Ignition Facility is the resultant center-of-mass motion of the imploding capsule. This residual velocity reduces the efficiency of converting laser energy into plasma temperature. A new analysis method extracts the effective hot spot motion by using information from multiple neutron time-of-flight (nToF) lines-of-sight (LoSs). This technique fits a near Gaussian spectrum to the nToF scope traces and overcomes reliance on models to relate the plasma temperature to the mean energy of the emitted neutrons. This method requires having at least four nToF LoSs. The results of this analysis will be compared to an approach where each LoS is analyzed separately and a model is used to infer the mean energy of the emitted neutrons.

One of the biggest challenges in high-energy physics is to analyze a complex mix of experimental and simulation data to gain new insights into the underlying physics. Currently, this analysis relies primarily on the intuition of trained experts often using nothing more sophisticated than default scatter plots. Many advanced analysis techniques are not easily accessible to scientists and not flexible enough to explore the potentially interesting hypotheses in an intuitive manner. Furthermore, results from individual techniques are often difficult to integrate, leading to a confusing patchwork of analysis snippets too cumbersome for data exploration. This paper presents a case study on how a combination of techniques from statistics, machine learning, topology, and visualization can have a significant impact in the field of inertial confinement fusion. We present the $$\mathrm{ND}^2\mathrm{AV}$$ND2AV: N-dimensional data analysis and visualization framework, a user-friendly tool aimed at exploiting the intuition and current workflow of the target users. The system integrates traditional analysis approaches such as dimension reduction and clustering with state-of-the-art techniques such as neighborhood graphs and topological analysis, and custom capabilities such as defining combined metrics on the fly. All components are linked into an interactive environment that enables an intuitive exploration of a wide variety of hypotheses while relating the results to concepts familiar to the users, such as scatter plots. $$\mathrm{ND}^2\mathrm{AV}$$ND2AV uses a modular design providing easy extensibility and customization for different applications. $$\mathrm{ND}^2\mathrm{AV}$$ND2AV is being actively used in the National Ignition Campaign and has already led to a number of unexpected discoveries.

This work details a model used to infer the 3-D structure of the stagnated hot-spot and shell of inertial confinement fusion implosion experiments at the National Ignition Facility. The model assumes that 3-D low-mode drive perturbations can account for the majority of stagnation asymmetries experimentally observed. It uses an adaptive sampling algorithm to navigate the 24-D input parameter space to find a 3-D x-ray flux asymmetry whose application to an otherwise symmetric implosion results in a consistent match between synthetic and experimental diagnostic observables. The model is applied to a series of experiments and is able to achieve a consistent match for over 41 different observables, providing a high-fidelity reconstruction of the stagnation hot-spot and shell profile.

The Bigfoot approach to indirect-drive inertial confinement fusion has been developed as a compromise trading high convergence and areal densities for high implosion velocities, large adiabats, and hydrodynamic stability. Shape control and predictability are maintained by using relatively short laser pulses and merging the shocks within the deuterium-tritium-ice layer. These design choices ultimately limit the theoretically achievable performance, and one strategy to increase the 1D performance is to reduce the shell adiabat by extending the pulse shape. However, this can result in the loss of low-mode symmetry control, as the hohlraum “bubble,” the high-Z material launched by the outer-cone beams during the early part of the laser pulse, has more time to expand and will eventually intercept inner-cone beams preventing them from reaching the hohlraum waist, thus losing an equatorial capsule drive. Experiments were performed to study the shape control and predictability with extended pulse shapes in Bigfoot implosions, reducing the adiabat from nominally α ∼ 4 to α ∼ 3 and otherwise very similar experimental parameters. The implosion shape was measured both in-flight and at stagnation, with near-round implosions and low levels of P2 asymmetry throughout, indicating a maintained symmetry control with extended pulse shapes.

Computer models of inertial confinement fusion (ICF) implosions play an essential role in experimental design and interpretation as well as our understanding of fundamental physics under the most extreme conditions that can be reached in the laboratory. Building truly predictive models is a significant challenge, with the potential to greatly accelerate progress to high yield and ignition. One path to more predictive models is to use experimental data to update the underlying physics in a way that can be extrapolated to new experiments and regimes. We describe a statistical framework for the calibration of ICF simulations using data collected at the National Ignition Facility (NIF). We perform Bayesian inferences for a series of laser shots using an approach that is designed to respect the physics simulation as much as possible and then build a second model that links the individual-shot inferences together. We show that this approach is able to match multiple X-ray and neutron diagnostics for a whole series of NIF “BigFoot” shots. Within the context of 2D radiation hydrodynamic simulations, our inference strongly favors a significant reduction in fuel compression over other known degradation mechanisms (namely, hohlraum issues and engineering perturbations). This analysis is expanded using a multifidelity technique to pick fuel-ablator mix from several candidate causes of the degraded fuel compression (including X-ray preheat and shock timing errors). Finally, we use our globally calibrated model to investigate the extra laser drive energy that would be required to overcome the inferred fuel compression issues in NIF BigFoot implosions.

Indirect drive converts high power laser light into x-rays using small high-Z cavities called hohlraums. X-rays generated at the hohlraum walls drive a capsule filled with deuterium–tritium (DT) fuel to fusion conditions. Recent experiments have produced fusion yields exceeding 50 kJ where alpha heating provides ~3× increase in yield over PdV work. Closing the gaps toward ignition is challenging, requiring optimization of the target/implosions and the laser to extract maximum energy. The US program has a three-pronged approach to maximize target performance, each closing some portion of the gap. The first item is optimizing the hohlraum to couple more energy to the capsule while maintaining symmetry control. Novel hohlraum designs are being pursued that enable a larger capsule to be driven symmetrically to both reduce 3D effects and increase energy coupled to the capsule. The second issue being addressed is capsule stability. Seeding of instabilities by the hardware used to mount the capsule and fill it with DT fuel remains a concern. Work reducing the impact of the DT fill tubes and novel capsule mounts is being pursed to reduce the effect of mix on the capsule implosions. There is also growing evidence native capsule seeds such as a micro-structure may be playing a role on limiting capsule performance and dedicated experiments are being developed to better understand the phenomenon. The last area of emphasis is the laser. As technology progresses and understanding of laser damage/mitigation advances, increasing the laser energy seems possible. This would increase the amount of energy available to couple to the capsule, and allow larger capsules, potentially increasing the hot spot pressure and confinement time. The combination of each of these focus areas has the potential to produce conditions to initiate thermo-nuclear ignition.

The National Ignition Campaign (NIC) on the National Ignition Facility plans to use an indirectly driven spherical implosion to assemble and ignite a mass of D-T fuel. The NIC is currently in the p...

We have developed and fielded x-ray penumbral imaging on the National Ignition Facility in order to enable sub-10 μm resolution imaging of stagnated plasma cores (hot spots) of spherically shock compressed spheres and shell implosion targets. By utilizing circular tungsten and tantalum apertures with diameters ranging from 20 μm to 2 mm, in combination with image plate and gated x-ray detectors as well as imaging magnifications ranging from 4 to 64, we have demonstrated high-resolution imaging of hot spot plasmas at x-ray energies above 5 keV. Here we give an overview of the experimental design criteria involved and demonstrate the most relevant influences on the reconstruction of x-ray penumbral images, as well as mitigation strategies of image degrading effects like over-exposed pixels, artifacts, and photon limited source emission. We describe experimental results showing the advantages of x-ray penumbral imaging over conventional Fraunhofer and photon limited pinhole imaging and showcase how internal hot spot microstructures can be resolved.

The electron temperature at stagnation of an ICF implosion can be measured from the emission spectrum of high-energy x-rays that pass through the cold material surrounding the hot stagnating core. Here we describe a platform developed on the National Ignition Facility where trace levels of a mid-Z dopant (krypton) are added to the fuel gas of a symcap (symmetry surrogate) implosion to allow for the use of x-ray spectroscopy of the krypton line emission.

The Neutron Imaging System at the National Ignition Facility is used to observe the primary ∼14 MeV neutrons from the hotspot and down-scattered neutrons (6-12 MeV) from the assembled shell. Due to the strong spatial dependence of the primary neutron fluence through the dense shell, the down-scattered image is convolved with the primary-neutron fluence much like a backlighter profile. Using a characteristic scattering angle assumption, we estimate the primary neutron fluence and compensate the down-scattered image, which reveals information about asymmetry that is otherwise difficult to extract without invoking complicated models.

An accurate understanding of burn dynamics in implosions of cryogenically layered deuterium (D) and tritium (T) filled capsules, obtained partly through precision diagnosis of these experiments, is essential for assessing the impediments to achieving ignition at the National Ignition Facility. We present measurements of neutrons from such implosions. The apparent ion temperatures T_{ion} are inferred from the variance of the primary neutron spectrum. Consistently higher DT than DD T_{ion} are observed and the difference is seen to increase with increasing apparent DT T_{ion}. The line-of-sight rms variations of both DD and DT T_{ion} are small, ∼150eV, indicating an isotropic source. The DD neutron yields are consistently high relative to the DT neutron yields given the observed T_{ion}. Spatial and temporal variations of the DT temperature and density, DD-DT differential attenuation in the surrounding DT fuel, and fluid motion variations contribute to a DT T_{ion} greater than the DD T_{ion}, but are in a one-dimensional model insufficient to explain the data. We hypothesize that in a three-dimensional interpretation, these effects combined could explain the results.

We present the outcome of an idea market run for one of GE Energy's sub-businesses in July and August of 2006. GE Energy used this market to elicit and rank-order technology and product ideas from across the sub-business. In this experiment, we examine the behavior of traders that have submitted the ideas on the market and their influence on the market's outcome. An idea’s submitter is clearly motivated to have his idea valued highly by the market, both by the funding given to the top idea as well as smaller prizes given to the top three ideas. In general, founders tended to buy their suggested ideas at prices above the volume-weighted-average-price (VWAP) in significant volumes. We discuss the implications and mitigation strategies. A survey of market participants yielded mixed results regarding the market's effectiveness at ranking ideas but very positive results regarding the quality of ideas proposed.

Targets intended to produce ignition on the National Ignition Facility (NIF) are being simulated, and the simulations are used to set specifications for target fabrication. Recent design work has focused on incorporating the implications of NIF experiments that were done in fall 2009 and planning for the campaign in 2010. Near-term experiments will use Ge-doped CH, although Be and diamond are still under active consideration for 2011 and beyond. The emphasis in this paper will be on changes in the requirements over the last year, the characteristics of the 2010 CH-ablator design, and the designs for 2011 and beyond. Capsule defects of particular interest are surface perturbations on the CH ablator and composition variations in the Be shells. Complete tables of specifications are regularly updated for all of the targets. All the specifications are rolled together into an error budget indicating adequate margin for ignition with all of the designs.

Machine learning is finding increasingly broad applications in the physical sciences. This most often involves building a model relationship between a dependent, measurable output, and an associated set of controllable, but complicated, independent inputs. We present a tutorial on current techniques in machine learning—a jumping-off point for interested researchers to advance their work. We focus on deep neural networks with an emphasis on demystifying deep learning. We begin with background ideas in machine learning and some example applications from current research in plasma physics. We discuss supervised learning techniques for modeling complicated functions, beginning with familiar regression schemes, and then advancing to more sophisticated deep learning methods. We also address unsupervised learning and techniques for reducing the dimensionality of input spaces. Along the way, we describe methods for practitioners to help ensure that their models generalize from their training data to as-yet-unseen test data. We describe classes of tasks—predicting scalars, handling images, and fitting time-series—and prepare the reader to choose an appropriate technique. We finally point out some limitations to modern machine learning and speculate on some ways that practitioners from the physical sciences may be particularly suited to help.

Targets intended to produce ignition on the National Ignition Facility are being simulated, and the simulations are used to set specifications for target fabrication. Recent design work has focused...

We describe a method for lock-in amplification that uses a chaotic reference signal and a synchronized receiver. This technique is compared with conventional lock-in methods using a periodic reference signal. The apparatus can be constructed in an undergraduate electronics lab using a light emitting diode for the transmitter and a photodiode for the receiver, and allows students to explore a novel measurement technique based on chaotic dynamics.