SEARCH · Engineering Papers
Results for “Variational inference”
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.
Posterior Covariance Matrix Approximations
Here, the Davis equation of state (EOS) is commonly used to model thermodynamic relationships for high explosive (HE) reactants. Typically, the parameters in the EOS are calibrated, with uncertainty, using a Bayesian framework and Markov Chain Monte Carlo (MCMC) methods. However, MCMC methods are computationally expensive, especially for complex models with many parameters. This paper provides a comparison between MCMC and less computationally expensive Variational methods (Variational Bayesian and Hessian Variational Bayesian) for computing the posterior distribution and approximating the posterior covariance matrix based on heterogeneous experimental data. All three methods recover similar posterior distributions and posterior covariance matrices. This study demonstrates that for this EOS parameter calibration application, the assumptions made in the two Variational methods significantly reduce the computational cost but do not substantially change the results compared to MCMC.
CONCURRENT, CONDENSED STEIN VARIATIONAL GRADIENT DESCENT FOR UNCERTAINTY QUANTIFICATION OF NEURAL NETWORKS
In this work, we propose a Stein variational gradient descent (SVGD) method to concurrently sparsify, train, and provide uncertainty quantification (UQ) of a complexly parameterized model, such as a neural network (NN). It employs a graph reconciliation and condensation process to reduce complexity and increase similarity in the Stein ensemble of parameterizations. Therefore, the proposed concurrent, condensed SVGD (ccSVGD) method can provide UQ on parameters, not just outputs. Furthermore, the parameter reduction speeds up the convergence of the Stein gradient descent as it reduces the combinatorial complexity by aligning and differentiating the sensitivity to parameters. These properties are demonstrated with an illustrative example and an application to a mechanical response representation problem in solid mechanics.
We Know Where We Don't Know: 3D Bayesian CNNs for Credible Geometric Uncertainty
Deep learning has been successfully applied to the segmentation of 3D Computed Tomography (CT) scans. Establishing the credibility of these segmentations requires uncertainty quantification (UQ) to identify untrustworthy predictions. Recent UQ architectures include Monte Carlo dropout networks (MCDNs), which approximate deep Gaussian processes, and Bayesian neural networks (BNNs), which learn the distribution of the weight space. BNNs are advantageous over MCDNs for UQ but are thought to be computationally infeasible in high dimension, and neither architecture has produced interpretable geometric uncertainty maps. We propose a novel 3D Bayesian convolutional neural network (BCNN), the first deep learning method which generates statistically credible geometric uncertainty maps and scales for application to 3D data. We present experimental results on CT scans of graphite electrodes and laser-welded metals and show that our BCNN outperforms an MCDN in recent uncertainty metrics. The geometric uncertainty maps generated by our BCNN capture distributions of sigmoid values that are interpretable as confidence intervals, critical for applications that rely on deep learning for high-consequence decisions.
Xenon in the Protoplanetary Disk (PPD), in Two Planets, and a Comet
Isotopic anomalies in several elements, as recently observed in meteorites, are generally interpreted to indicate nonequilibrium environments in the protoplanetary disk (PPD). Here we study reported Xe isotopic compositions on planets Earth and Mars, in a comet, and in meteorites for precursor discrepancies. Abundance variations of inferred presolar nano-diamonds, the carrier phase of the Xe-HL component, appear to be the primary source of nonuniformity of Xe precursors in the PPD, together with mechanisms of mass-dependent fractionation. While planet Mars kept a record of initial solar Xe isotopic abundances, such a record is missing for planet Earth. Xe isotopic abundances in paleo-atmospheres of both planets represent secondary reservoirs that show mass-dependent fractionation effects, but the inferred compositions of their PPD precursors differ: Mars atmospheric precursor Xe had solar isotopic composition, while Earth’s Xe precursor is consistent with a PPD reservoir of low nano-diamond abundance. Strong mass-dependent fractionation effects are observed in Xe components of IAB irons and in Yamato carbonaceous (CY) chondrites, and show that fractionation mechanisms are not restricted to planetary atmospheres. These records show that Xe isotopes in solar system reservoirs are useful tracers of evolutionary processes and of nonequilibrated presolar components in the PPD.
SQUID Readout of a High-$Q$ Superconducting $LC$ Resonator
We demonstrate the readout of a superconducting $LC$ resonator, with $Q \approx 2 \times 10^{6}$ at approximately 250 kHz, using a dc SQUID followed by a SQUID series-array amplifier readout chain. We find that $Q$ depends on the SQUID flux-bias point, increasing near the shallow-slope point of the SQUID modulation curve and decreasing near the steep-slope point, consistent with the previously observed SQUID damping effects. From this variation, we infer the SQUID effective input impedance. We further infer the effective temperature of the resonator circuit from the resonance peak in the noise spectrum and show that it also depends on the SQUID flux-bias point, suggesting a possible contribution from SQUID back-action noise. This system provides a working prototype for future experiments based on lumped-element resonators with SQUID readout, in particular low-mass axion searches within the DMRadio program.
Compositional characteristics of some Apollo 14 clastic materials.
Eighty-two subsamples of Apollo 14 materials have been analyzed by instrumental neutron activation analysis techniques for as many as 25 elements. In many cases, it was necessary to develop new procedures to allow analyses of small specimens. Compositional relationships among Apollo 14 materials indicate that there are small but systematic differences between regolith from the valley terrain and that from Cone Crater ejecta. Fragments from 1-2 mm size fractions of regolith samples may be divided into compositional classes, and the 'soil breccias' among them are very similar to valley soils. Multicomponent linear mixing models have been used as interpretive tools in dealing with data on regolith fractions and subsamples from breccia 14321. These mixing models show systematic compositional variations with inferred age for Apollo 14 clastic materials.
Ion chemistry of N2/+/ and the solar ultraviolet flux in the thermosphere
We have analyzed daytime composition measurements in the thermosphere from Atmosphere Explorer-C by use of a theory of the ion chemistry. Predicted and measured N2(+) densities are compared to infer global variation of the solar ionizing flux. Results indicate that the solar flux of Hinteregger measured on Atmosphere Explorer-C provides an adequate basis for the calculation of solar ultraviolet photoionization rates. The role of metastable species is examined, and theoretical O+(2D) densities are presented.
On the supposed anticorrelation of solar polar and equatorial rotation rates
The anticorrelation between two of the three parameters used to calculate the angular velocity of the sun's differential rotation is here said to be due to numerical coupling. A computer simulation technique shows that the relationship between the two parameters is caused by the effect of noise on the least-squares analysis used to obtain the three parameters used to determine the angular velocity in terms of the heliographic latitude. The computer simulation technique is described. The supposed anticorrelation had been used to infer that variations of the sun's polar and equatorial rotation rates are anticorrelated.
Alteration in CM carbonaceous chondrites inferred from modal and chemical variations in matrix
The modal abundance of matrix in CM chondrites appears to vary from 57-85 vol%. The concentrations of volatile elements that should occur in matrix remain approximately constant despite differences in the proportions of matrix, suggesting that the differing matrix contents may not be real primary variations but are optical effects due to aqueous alteration processes that make other petrologic components unrecognizable. Apparent matrix content can be used as a qualitative measure of the degree of alteration experienced by each CM chondrite. Fe/Si ratios in matrices decrease progressively with increasing alteration due to the formation of new phyllosilicate phases with higher Mg/Fe ratios and optically recognizable opaque minerals that are not counted as matrix. The aqueous alteration process in CM chondrites appears to have been largely isochemical if the bulk meteorites are considered as the reacting systems, although depletion patterns and isotopic anomalies indicate open-system behavior for a few highly mobile components.
Monitoring snowpack properties by passive microwave sensors on board of aircraft and satellites
Snowpack properties such as water equivalent and snow wetness may be inferred from variations in measured microwave brightness temperatures. This is because the emerged microwave radiation interacts directly with snow crystals within the snowpack. Using vertically and horizontally polarized brightness temperatures obtained from the multifrequency microwave radiometer (MFMR) on board a NASA research aircraft and the electrical scanning microwave radiometer (ESMR) and scanning multichannel microwave radiometer (SMMR) on board the Nimbus 5, 6, and 7 satellites, linear relationships between snow depth or water equivalent and microwave brightness temperature were developed. The presence of melt water in the snowpack generally increases the brightness temperatures, which can be used to predict snowpack priming and timing of runoff.
11-year cycle solar modulation of cosmic ray intensity inferred from C-14 content variation in dated tree rings
A liquid scintillation-photomultiplier tube counter system was used to measure the Delta-C-14 values of 60 tree rings, dating from 1866 to 1925, that were taken from a white spruce grown in Canada at 68 deg N, 130 deg W. A 10-percent variation is found which is anticorrelated with sunspot numbers, although the amplitude of the variation is 2-3 times higher than expected in trees grown at lower latitudes. A large dip in the data at about 1875 suggests an anomalously large modulation of cosmic ray intensity during the 1867-1878 AD solar cycle, which was the most active of the 19th century.
Evidence for a thermospheric source of stratospheric NOx
Analyses of a three-year time series of rocket ozone measurements at Wallops Island, VA, a set of rocket ozone soundings across the Southern Hemisphere, and rocket soundings at Fort Churchill, Manitoba are reported. Evidence is obtained that the NOx budget is not simply explained by oxidation of biospheric nitrous oxide. I 1-D time-dependent photochemical model is used to compute the amount of NO2 required to maintain odd oxygen in a steady state after accounting for Chapman, odd hydrogen, and odd chlorine reactions. At Wallops Island, a mid-latitude station, the inferred seasonal variation of NOx is small with the fall and winter mixing ratios about 20 percent greater than the spring and summer values. The soundings at Fort Churchill require about the same NOx amount as at Wallops Island in the spring and summer months but more than twice this amount in late fall and winter. Results indicate that the nitrous oxide source of NOx is supplemented by a polar source during the fall and winter months. This is consistent with the descent of thermospheric air with its high nitric oxide content during the period of strong cooling in the polar night.
Velocity-resolved observations of water in comet Halley
High-resolution observations of H2O in comet Halley are presented. The positions and widths of the H2O emission line profiles are indicative of the coma's velocity field characteristics. Spectral line shape changes are used to infer temporal variations in the kinematic properties of the outflow. The results suggest that H2O is released into the coma by way of multiple jets.
Consequences of experimental transient rheology
Recent analyses of transient creep data for lower crustal and upper mantle substances are used to constrain the viscosity contrast across the 670 km discontinuity, and a 4-layer earth model in which both upper and lower mantles are described by a Burgers body rheology is assumed. Results indicate that the previously inferred viscosity variations would be reduced by a factor of ten using the new transient models. This result is in agreement with a revised viscosity estimate based on long-wavelength geoid anomalies and seismic tomography. Time-dependent perturbations to the gravity field from recent ice movements are shown to be significant.
Rapid mass-loss transients in VV Cephei
Biweekly ultraviolet observations of the red supergiant-hot dwarf binary VV Cephei during 1991, obtained near third quadrature, have revealed the existence of short-term continuum variations. We infer these are superposed on an underlying emission-line spectrum. The viewing geometry of this long-period system suggests we are seeing a process associated with nonuniform mass transfer to an accretion disk. This rapid variability can be related to global instabilities in the stellar wind and mass loss from the red supergiant.
Lightning Return-Stroke Current Waveforms Aloft, from Measured Field Change, Current, and Channel Geometry
Three-dimensional reconstructions of six rocket-triggered lightning channels are derived from stereo photographs. These reconstructed channels are used to infer the behavior of the current in return strokes above the ground from current waveforms measured at the channel base and electric-field-change waveforms measured at a range of 5.2 kilometers for 24 return strokes in these channels. Streak photographs of 14 of the same strokes are analyzed to determine the rise times, propagation speeds, and amplitudes of relative light intensity for comparison with the electrical inferences. Results include the following: 1) The fine structure of the field-change waveforms that were radiated by these subsequent return strokes can be explained, in large part, by channel geometry. 2) The average 10 - 90% rise time of the stroke current increased by about a factor of seven in our sample, from an observed 0.31 plus or minus 0.17 microseconds at the surface to an inferred 2.2 plus or minus 0.5 microcseconds at 1 kilometer path length above the surface. 3) The three-dimensional propagation speed of the current front averaged 1.80 plus or minus 0.24 X 10(exp 8) meters per second over channel lengths typically greater than 1 kilometer. 4) Assuming that the measured current was entirely due to the return stroke forced an unreasonably large and abrupt reduction in inferred current amplitude over the first few tens of meters above the surface, especially in cases when the leader was bright relative to its stroke. Therefore, a significant fraction of the current at the surface was probably due to the leader, at least in such cases. 5) Peak return-stroke currents decreased by approximately 37 plus or minus 12% from 100 meters to 1 kilometer of path length above the surface. Because of uncertainty about how to partition the measured current between leader and return stroke, we are unable to infer the variation of current amplitude near the ground.
Variability in the Speed of the Brewer-Dobson Circulation as Observed by Aura/MLS
We use Aura/MLS stratospheric water vapour (H2O) measurements as tracer for dynamics and infer interannual variations in the speed of the Brewer-Dobson circulation (BDC) from 2004 to 2011. We correlate one-year time series of H2O in the lower stratosphere at two subsequent pressure levels (68 hPa, approx.18.8 km and 56 hPa, approx 19.9 km at the Equator) and determine the time lag for best correlation. The same calculation is made on the horizontal on the 100 hPa (approx 16.6 km) level by correlating the H2O time series at the Equator with the ones at 40 N and 40 S. From these lag coefficients we derive the vertical and horizontal speeds of the BDC in the tropics and extra-tropics, respectively. We observe a clear interannual variability of the vertical and horizontal branch. The variability reflects signatures of the Quasi Biennial Oscillation (QBO). Our measurements confirm the QBO meridional circulation anomalies and show that the speed variations in the two branches of the BDC are out of phase and fairly well anti-correlated. Maximum ascent rates are found during the QBO easterly phase. We also find that transport of H2O towards the Northern Hemisphere (NH) is on the average two times faster than to the Southern Hemisphere (SH) with a mean speed of 1.15m/s at 100 hPa. Furthermore, the speed towards the NH shows much more interannual variability with an amplitude of about 21% whilst the speed towards the SH varies by only 10 %. An amplitude of 21% is also observed in the variability of the ascent rate at the Equator which is on the average 0.2mm/s.