Search NASA⌕ Search

SEARCH · Search NASA

Results for “distributed estimation”

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 1,135 records · Page 63

A Graph Convolutional Network for Active Distribution System Anomaly Detection Considering Measurement Spatial-Temporal Correlations

The accuracy of distribution system state estimation may be significantly impacted by the existence of bad measure-ments and unexpected topology errors. This paper proposes a data-driven Graph Convolutional Network (GCN) for anomaly detection, including bad measurements and topology change events. Compared to many existing machine learning approaches, the proposed approach embeds both spatial-temporal measure-ment correlations, which allows us to detect and distinguish different anomalies. Numerical results carried out on the IEEE 37-node system demonstrate that the proposed-based method can obtain high accuracy in detecting bad data and topology changes as compared to other approaches, even in the presence of high PV penetrations.

active distribution system↗

Data-driven, Interpretable Photometric Redshifts Trained on Heterogeneous and Unrepresentative Data

We present a new method for inferring photometric redshifts in deep galaxy and quasar surveys, based on a data-driven model of latent spectral energy distributions (SEDs) and a physical model of photometric fluxes as a function of redshift. This conceptually novel approach combines the advantages of both machine learning methods and template fitting methods by building template SEDs directly from the spectroscopic training data. This is made computationally tractable with Gaussian processes operating in flux–redshift space, encoding the physics of redshifts and the projection of galaxy SEDs onto photometric bandpasses. This method alleviates the need to acquire representative training data or to construct detailed galaxy SED models; it requires only that the photometric bandpasses and calibrations be known or have parameterized unknowns. The training data can consist of a combination of spectroscopic and deep many-band photometric data with reliable redshifts, which do not need to entirely spatially overlap with the target survey of interest or even involve the same photometric bands. We showcase the method on the i -magnitude-selected, spectroscopically confirmed galaxies in the COSMOS field. The model is trained on the deepest bands (from SUBARU and HST ) and photometric redshifts are derived using the shallower SDSS optical bands only. We demonstrate that we obtain accurate redshift point estimates and probability distributions despite the training and target sets having very different redshift distributions, noise properties, and even photometric bands. Our model can also be used to predict missing photometric fluxes or to simulate populations of galaxies with realistic fluxes and redshifts, for example.

79 ASTRONOMY AND ASTROPHYSICS↗

Decentralized Microgrid Protection Through Relative Fault Direction Classification: Preprint

Protection in inverter-based resources (IBRs) dominated microgrids generally face significant challenges due to the low fault current and inconsistent fault behaviors from IBRs. Recently, machine learning-based approaches have attracted considerable attention to address these challenges. This paper introduces a novel decentralized protection strategy for microgrids. The proposed method decomposes the protection challenge into several distributed learning tasks, enabling individual relays to autonomously determine the direction of faults using a binary classification framework based on support vector machine (SVM) algorithms. Following the distributed fault direction estimation, classifier outcomes are shared among neighboring relays, facilitating a local decision-making process to ascertain the presence of faults within the neighborhood. Finally, a tripping signal is generated based on the classifier results of each relay to operate the circuit breaker. To test and validate this approach, a 100% renewable microgrid model is simulated in MATLAB/Simulink. In the numerical analysis, the application of SVM classifiers in our approach yields impressive results: an average relay classification accuracy of 98%, and a 96% accuracy in circuit breaker control. These findings highlight the potential of machine-learning-based approaches in enhancing the efficiency and reliability of microgrid protection systems.

decentralized algorithm↗

SU-E-T-422: Fast Analytical Beamlet Optimization for Volumetric Intensity-Modulated Arc Therapy

Purpose: To implement a fast optimization algorithm on CPU/GPU heterogeneous computing platform and to obtain an optimal fluence for a given target dose distribution from the pre-calculated beamlets in an analytical approach. Methods: The 2D target dose distribution was modeled as an n-dimensional vector and estimated by a linear combination of independent basis vectors. The basis set was composed of the pre-calculated beamlet dose distributions at every 6 degrees of gantry angle and the cost function was set as the magnitude square of the vector difference between the target and the estimated dose distribution. The optimal weighting of the basis, which corresponds to the optimal fluence, was obtained analytically by the least square method. Those basis vectors with a positive weighting were selected for entering into the next level of optimization. Totally, 7 levels of optimization were implemented in the study.Ten head-and-neck and ten prostate carcinoma cases were selected for the study and mapped to a round water phantom with a diameter of 20cm. The Matlab computation was performed in a heterogeneous programming environment with Intel i7 CPU and NVIDIA Geforce 840M GPU. Results: In all selected cases, the estimated dose distribution was in a good agreement with the given target dose distribution and their correlation coefficients were found to be in the range of 0.9992 to 0.9997. Their root-mean-square error was monotonically decreasing and converging after 7 cycles of optimization. The computation took only about 10 seconds and the optimal fluence maps at each gantry angle throughout an arc were quickly obtained. Conclusion: An analytical approach is derived for finding the optimal fluence for a given target dose distribution and a fast optimization algorithm implemented on the CPU/GPU heterogeneous computing environment greatly reduces the optimization time.

60 APPLIED LIFE SCIENCES↗

Measuring Z{sup '} couplings at the CERN LHC

We study the properties of potential new Z{sup '} gauge bosons produced through the Drell-Yan mechanism at the Large Hadron Collider (LHC). Our analysis is performed using a fully differential next-to-leading-order QCD calculation with spin correlations, interference effects, and experimental acceptances included. We examine the distinguishability of different models and the feasibility of extracting general coupling information with statistical, residual scale, and current parton distribution function error estimates included. We extend a previous parametrization of Z{sup '} couplings to include parity-violating coupling combinations and introduce a convenient technique for simulating new gauge bosons on-peak using the concept of basis models. We illustrate our procedure using several example Z{sup '} models. We find that one can extract reliably four combinations of generation-independent quark and lepton couplings in our analysis. For a Z{sup '} mass of 1.5 TeV, one can determine coupling information very well assuming 100 fb{sup -1} of integrated luminosity, and a precise measurement becomes possible with 1ab{sup -1} at the super-LHC (SLHC). For a 3 TeV mass, a reasonable determination requires the SLHC.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Structural and Stratigraphic Controls on Methane Hydrate occurrence and distribution: Gulf of Mexico, Walker Ridge 313 and Green Canyon 955: Final Report

The goal of this project was to determine structural and stratigraphic controls on hydrate occurrence and distribution in Green Canyon (GC) 955 and Walker Ridge (WR) 313 blocks using seismic and well data. Gas hydrate was discovered in these blocks in coarse- and fine-grained sediments during the 2009 Joint Industrial project (JIP) Leg 11 drilling expedition. Although the immediate interest of the exploration community is exclusively hydrate which is present in coarse–grained sediments, factors that control hydrate and free gas distribution in the two blocks and whether coarse and fine-grained hydrate-bearing units are related in any manner, formed the core of this research. The project spanned from 10/01/2012 to 07/31/2016. In the project, in both the leased blocks, the interval spanning the gas hydrate stability zone (GHSZ) was characterized using a joint analysis of sparse Ocean Bottom Seismic (OBS) and dense, surface–towed multichannel seismic (MCS) data. The project team had the luxury of calibrating their results with two well logs. Advance processing methods such as depth migration and full-waveform inversion (FWI) were used for seismic data analysis. Hydrate quantification was achieved through interpretation of the FWI velocity field using appropriate rock physics models at both blocks. The seismic modeling/inversion methodology (common to both GC955 and WR313 blocks) was as follows. First, the MCS data were depth migrated using a P-wave velocity (VP) model constructed using inversion of reflection arrival times of a few (four in both cases) key horizons carefully picked in the OBS data to farthest possible offsets. Then, the resolution of the traveltime VP model was improved to wavelength scale by inverting OBS gathers up to the highest frequency possible (21.75 Hz for GC955 and 17.5 for WR313) using FWI. Finally, the hydrate saturation (or the volume fraction) was estimated at the well location assuming one of the other hydrate morphology (filling the primary or the secondary porosity) was extrapolated out from the wells using the FWI VP as a guide. General outcomes were as follows. First and foremost, an imaging methodology using sparse seismic data, which is easily replicable at other sites with similar datasets, has been demonstrated. The end product of this methodology at both the leased blocks is quantitative estimates of hydrate distribution. Second, at both locations there is strong evidence that the base of the GHSZ, which does not appear as a clear Bottom Simulating Reflection (BSR), manifests in the VP perturbations created by FWI, suggesting that FWI is sensitive to subtle compositional changes in shallow sediments and establishes it as a valuable tool for investigations of hydrate-bearing basins. Third, through joint interpretation of the depth migrated image and the FWI VP model, how structure and stratigraphy jointly determine hydrate and free gas distribution in both blocks could be clearly visualized. The joint interpretation also suggests that the coarse and fine grained hydrate-bearing sediments at both leased are connected. Site specific results, in addition to general results, are as follows. At GC955 the overlying fine-grained hydrate-bearing unit could have been sourced from the underlying hydrate coarse-grained channel-levee complex through a chimney feature. The channel-levee system at GC955 is compartmentalized by faults, of which only a few may be impermeable. Although compartmentalized, the channel-levee system in the GC955 as a whole might be in communication except selected zones. At WR313 the overlying fine-grained fracture-filled hydrate unit appears to be sourced from below the GHSZ. The reason that only a particular fine-grained unit has hydrate, despite having lower porosity that the bounding units, could be the presence of secondary porosity (such as those formed from clay dewatering under compaction). In conclusion, the project was a pioneering effort in in joint analysis of OBS and MCS datasets for advancing the knowledge about a hydrate and free–gas system dynamics using advanced processing methods such as FWI and depth migration. Results obtained in this project can greatly advance the tools and techniques used for delineating specific hydrate prospects. Results obtained in this project can also be seamlessly incorporated into other DOE funded project on modeling the potential productivity and commercial viability of hydrate from sand-dominated reservoirs. The OBS and MCS data in this project were acquired in 2012 (after the JIP II drilling) by the USGS and therefore the results are a posteriori. Nonetheless, the seismic inversion workflow established through this project can be used to generate various what-if quantification scenarios even in absence of logs and serve as a valuable tool for guiding drilling operations. Results from this project can augment other DOE sponsored projects on determining the commercial viability of methane production from the Gulf of Mexico.

03 NATURAL GAS↗

Exploring the Mass-loss Histories of the Red Supergiants

We report mid- to far-infrared imaging and photometry from 7 to 37 μm with SOFIA/FORCAST and 2 μm adaptive optics imaging with LBTI/LMIRCam of a large sample of red supergiants (RSGs) in four Galactic clusters: RSGC1, RSGC2 = Stephenson 2, RSGC3, and NGC 7419. The RSGs in these clusters cover their expected range in luminosity and initial mass from ≈9 to more than 25 M {sub ⊙}. The population includes examples of very late-type RSGs such as MY Cep, which may be near the end of the RSG stage, high-mass-losing maser sources, yellow hypergiants, and post-RSG candidates. Many of the stars and almost all of the most luminous have spectral energy distributions (SEDs) with extended infrared excess radiation at the longest wavelengths. To best model their SEDs, we use the DUSTY code with a variable radial density distribution function to estimate their mass-loss rates. Our M-dot –luminosity relation for 42 RSGs basically follows the classical de Jager curve, but at luminosities below 10{sup 5} L{sub ⊙}, we find a significant population of RSGs with M-dot below the de Jager relation. At luminosities above 10{sup 5} L {sub ⊙}, there is a rapid transition to higher mass-loss rates that approximates and overlaps the de Jager curve. We recommend that instead of using a linear relation or single curve, the empirical M-dot –luminosity relation is better represented by a broad band. Interestingly, the transition to much higher M-dot at about 10{sup 5} L {sub ⊙} corresponds approximately to an initial mass of 18–20 M {sub ⊙}, which is close to the upper limit for RSGs becoming Type II supernovae.

79 ASTRONOMY AND ASTROPHYSICS↗

Constraints on primordial non-Gaussianity from galaxy-CMB lensing cross correlation

Recent studies have shown that the primordial non-Gaussianity affects the clustering of dark matter halos through a scale-dependent bias, and various constraints on the non-Gaussianity through this scale-dependent bias have been placed. Here, we introduce the cross correlation between the CMB lensing potential and the galaxy angular distribution to effectively extract information about the bias from the galaxy distribution. Then, we estimate the error of the nonlinear parameter, f{sub NL}, for the ongoing CMB experiments and galaxy surveys, such as Planck and Hyper Suprime-Cam. We found that for the constraint on f{sub NL} with Planck and Hyper Suprime-Cam, the wide field galaxy survey is preferable to the deep one, and the expected error on f{sub NL} can be as small as {Delta}f{sub NL{approx}}80 for b{sub 0}=2 and {Delta}f{sub NL{approx}}30 for b{sub 0}=4, where b{sub 0} is the linear bias parameter. It is also found that a future wide field galaxy survey could achieve {Delta}f{sub NL{approx}}5 with cosmic microwave background prior to Planck, if one could observe highly biased objects at higher redshift (z{approx}2).

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Dosimetric impact of an {sup 192}Ir brachytherapy source cable length modeled using a grid-based Boltzmann transport equation solver

Purpose: To evaluate the dose distributions of an {sup 192}Ir source (model VS2000) in homogeneous water geometry calculated using a deterministic grid-based Boltzmann transport equation solver (GBBS) in the commercial treatment planning system (TPS) (BRACHYVISION-ACUROS v8.8). Methods: Using percent dose differences (%{Delta}D), the GBBS (BV-ACUROS) was compared to the (1) published TG-43 data, (2) MCNPX Monte Carlo (MC) simulations of the {sup 192}Ir source centered in a 15 cm radius water sphere, and (3) TG-43 output from the TPS using vendor supplied (BV-TG43-vendor) and user extended (BV-TG43-extended) 2D anisotropy functions F(r,{theta}). BV-ACUROS assumes 1 mm of NiTi cable, while the TPS TG-43 algorithm uses data based on a 15 cm cable. MC models of various cable lengths were simulated. Results: The MC simulations resulted in >20% dose deviations along the cable for 1, 2, and 3 mm cable lengths relative to 15 cm. BV-ACUROS comparisons with BV-TG43-vendor and BV-TG43-extended yielded magnitude of differences, consistent with those seen in MC simulations. However, differences >20% extended further ({theta}{<=}10 deg.) when using the vendor supplied anisotropy function F{sub ven}(r,{theta}). These differences were also seen in comparisons of F(r,{theta}) derived from the TPS output. Conclusions: The results suggest that %{Delta}D near the cable region is larger than previously estimated. The spatial distribution of the dose deviation is highly dependent on the reference TG-43 data used to compare to GBBS. The differences observed, while important to realize, should not have an impact on clinical dosimetry in homogeneous water.

61 RADIATION PROTECTION AND DOSIMETRY↗

Estimation of end of life mobile phones generation: The case study of the Czech Republic

Highlights: Black-Right-Pointing-Pointer In this paper, we define lifespan of mobile phones and estimate their average total lifespan. Black-Right-Pointing-Pointer The estimation of lifespan distribution is based on large sample of EoL mobile phones. Black-Right-Pointing-Pointer Total lifespan of Czech mobile phones is surprisingly long, exactly 7.99 years. Black-Right-Pointing-Pointer In the years 2010-20, about 26.3 million pieces of EoL mobile phones will be generated in the Czech Republic. - Abstract: The volume of waste electrical and electronic equipment (WEEE) has been rapidly growing in recent years. In the European Union (EU), legislation promoting the collection and recycling of WEEE has been in force since the year 2003. Yet, both current and recently suggested collection targets for WEEE are completely ineffective when it comes to collection and recycling of small WEEE (s-WEEE), with mobile phones as a typical example. Mobile phones are the most sold EEE and at the same time one of appliances with the lowest collection rate. To improve this situation, it is necessary to assess the amount of generated end of life (EoL) mobile phones as precisely as possible. This paper presents a method of assessment of EoL mobile phones generation based on delay model. Within the scope of this paper, the method has been applied on the Czech Republic data. However, this method can be applied also to other EoL appliances in or outside the Czech Republic. Our results show that the average total lifespan of Czech mobile phones is surprisingly long, exactly 7.99 years. We impute long lifespan particularly to a storage time of EoL mobile phones at households, estimated to be 4.35 years. In the years 1990-2000, only 45 thousands of EoL mobile phones were generated in the Czech Republic, while in the years 2000-2010 the number grew to 6.5 million pieces and it is estimated that in the years 2010-2020 about 26.3 million pieces will be generated. Current European legislation sets targets on collection and recycling of WEEE in general, but no specific collection target for EoL mobile phone exists. In the year 2010 only about 3-6% of Czech EoL mobile phones were collected for recovery and recycling. If we make similar estimation using an estimated average EU value, then within the next 10 years about 1.3 billion of EoL mobile phones would be available for recycling in the EU. This amount contains about 31 tonnes of gold and 325 tonnes of silver. Since Europe is dependent on import of many raw materials, efficient recycling of EoL products could help reduce this dependence. To set a working system of collection, it will be necessary to set new and realistic collection targets.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Dynamics of the reaction C({sup 3}P)+SiH{sub 4}: Experiments and calculations

We conducted the reaction C({sup 3}P)+SiH{sub 4} at a collision energy of 4.0 kcal mol{sup -1} in a crossed molecular-beam apparatus measuring time-of-flight mass spectra and selective photoionization. Product ions with m/z=41-43 are associated with two product channels, H{sub 2}SiCH/HSiCH{sub 2}/SiCH{sub 3}+H and H{sub 2}SiC/HSiCH/SiCH{sub 2}+H{sub 2}. Apart from daughter ions and isotopic variants of reaction products, the species observed at m/z=43 is assigned to product H{sub 2}SiCH/HSiCH{sub 2}/SiCH{sub 3} and that at m/z=42 to product H{sub 2}SiC/HSiCH/SiCH{sub 2}. The signals observed at m/z=41 are due to dissociative ionization of silicon-carbon hydrides of these two types. We report time-of-flight spectra of products at specific laboratory angles and theoretical simulations, from which both kinetic-energy and angular distributions of products in the center-of-mass frame were derived. The release of kinetic energy is weakly dependent on the scattering angle for these two reactions. The channels for loss of H and H{sub 2} release average translational energies of 10.5 and 16.7 kcal mol{sup -1}, respectively. As hydrogen transfer before decomposition is facile, products H{sub 2}SiCH/HSiCH{sub 2}/SiCH{sub 3} and H{sub 2}SiC/HSiCH/SiCH{sub 2} exhibit mildly forward/backward preferred and isotropic angular distributions, respectively. We estimate the branching ratios of these channels for loss of H and H{sub 2} to be roughly 6:4. The measurements of release of kinetic energy and ionization thresholds of products indicate that SiCH{sub 3}({sup 2}A{sup ''}) and SiCH{sub 2}({sup 3}A{sub 2}) are dominant among isomeric products. To explore the reaction mechanism, we computed the potential-energy surfaces for the reaction C({sup 3}P)+SiH{sub 4}. The most likely mechanism is that atom C {sup 3}P inserts into bond Si-H of SiH{sub 4} in the entrance channel, and the reaction complex H{sub 3}SiCH subsequently isomerizes to HSiCH{sub 3} followed by decomposition to SiCH{sub 3}({sup 2}A{sup ''})+H and SiCH{sub 2}({sup 3}A{sub 2})+H{sub 2}. We observed no significant evidence for the reaction C({sup 1}D)+SiH{sub 4}.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

A Monte Carlo Technique to Estimate Emissions from R&D Facilities

The Pacific Northwest National Laboratory (PNNL) operates multidisciplinary laboratory Research and Development (R&D) facilities for the U. S. Department of Energy and sampled air chemical emissions from some of these facilities from 1998 until 2008. Although sampling was not required for environmental compliance, PNNL conducted sampling to provide data to compare estimated release fractions to those used for emissions estimates and to verify that methods used to determine compliance with air regulations and permits conservatively predict actual emissions. Sampling also identifies and quantifies air toxics emitted to compare with compliance limits established by the State. The sampling data provides a unique opportunity to quantify emissions and investigate factors potentially impacting emissions. Previous work using this data set include a paper by Woodruff, Benar, and McCarthy1 who summarized the compliance approach used by PNNL and described sampling and analytical measurements for the first sampling campaigns. Conclusions reported in the paper were that none of the measurements of the target compounds exceeded a State acceptable source impact level (ASIL)2 even using significant overestimation factors, and that an average release fraction calculated from the data provided reasonable validation of the factor used in compliance assessments. Other reports using the data include the following: • A metric was developed by Ballinger, Duchsherer, and Metoyer3 to compare chemical signatures. The analysis found similarities in chemical signatures from three of the buildings but a markedly different signature from the fourth building. • A method was developed and used on the data to rank the chemical compounds that present the greatest risk to a potential downstream receptor and to determine whether the sampling parameters and detection limits provided sufficient resolution to verify compliance at potential receptor locations.4 • The stack data from one of the buildings was analyzed using a statistical technique called Positive Matrix Factorization (PMF) to identify the number and composition of major contributing sources to measured stack concentrations.5 In this analysis, a method is described that uses a Monte Carlo technique to produce a distribution of emission estimates from the measured data. The method is applied to one of the chemicals previously identified as more significant than others using the ranking process.4 Variations on the method are investigated and compared to determine their effects on the emission estimates.

Laboratory, air chemical emissions, volatile organ↗

RESOLVING THE ACTIVE GALACTIC NUCLEUS AND HOST EMISSION IN THE MID-INFRARED USING A MODEL-INDEPENDENT SPECTRAL DECOMPOSITION

We present results on the spectral decomposition of 118 Spitzer Infrared Spectrograph (IRS) spectra from local active galactic nuclei (AGNs) using a large set of Spitzer/IRS spectra as templates. The templates are themselves IRS spectra from extreme cases where a single physical component (stellar, interstellar, or AGN) completely dominates the integrated mid-infrared emission. We show that a linear combination of one template for each physical component reproduces the observed IRS spectra of AGN hosts with unprecedented fidelity for a template fitting method with no need to model extinction separately. We use full probability distribution functions to estimate expectation values and uncertainties for observables, and find that the decomposition results are robust against degeneracies. Furthermore, we compare the AGN spectra derived from the spectral decomposition with sub-arcsecond resolution nuclear photometry and spectroscopy from ground-based observations. We find that the AGN component derived from the decomposition closely matches the nuclear spectrum with a 1σ dispersion of 0.12 dex in luminosity and typical uncertainties of ∼0.19 in the spectral index and ∼0.1 in the silicate strength. We conclude that the emission from the host galaxy can be reliably removed from the IRS spectra of AGNs. This allows for unbiased studies of the AGN emission in intermediate- and high-redshift galaxies—currently inaccesible to ground-based observations—with archival Spitzer/IRS data and in the future with the Mid-InfraRed Instrument of the James Webb Space Telescope. The decomposition code and templates are available at http://denebola.org/ahc/deblendIRS.

79 ASTRONOMY AND ASTROPHYSICS↗

Representative elementary volume estimation for porosity, moisture saturation, and air-water interfacial areas in unsaturated porous media: Data quality implications

Achieving a representative elementary volume (REV) has become a de facto criterion for demonstrating the quality of {mu}CT measurements in porous media systems. However, the data quality implications of an REV requirement have not been previously examined. In this work, deterministic REVs for porosity, moisture saturation (S{sub W}), and air-water interfacial area (A{sub I}) were estimated using a set of 49 {mu}CT images of eight unsaturated homogeneous porous media with heterogeneity in moisture distributions present in varying degrees. Estimated porosity REVs were <8 mm{sup 3} for all cases, smaller than typical CT image sizes ({approx}100 mm{sup 3}). Estimated S{sub W} and A{sub I} REVs were <55 mm{sup 3} for cases with homogeneous moisture distributions but could not be estimated for cases with heterogeneous moisture distributions, due to the absence of a distinct 'REV plateau' within the maximum imaged volume. Conventionally, S{sub W} and A{sub I} data from such non-REV cases would be excluded. The implications of excluding data on the basis of REV were examined by comparing A{sub I}-S{sub W} data measured on image windows of increasing size against the expected linear A{sub I}-S{sub W} relationship. At measurement scales exceeding porosity REV, random fluctuations in A{sub I}-S{sub W} data were excluded, even for cases containing heterogeneous moisture distributions. In contrast, requiring measurement scales to exceed S{sub W} and A{sub I} REV appeared overly restrictive and resulted in visible loss of reliable A{sub I}-S{sub W} data. We attribute these findings to overestimation of REVs due to inherently problematic estimation of deterministic REVs in real systems. Implications of these findings for ensuring CT data quality and the efficient use of CT data are discussed.

99 GENERAL AND MISCELLANEOUS↗

Joint Bayesian Estimation of Quasar Continua and the Lyα Forest Flux Probability Distribution Function

We present a new Bayesian algorithm making use of Markov Chain Monte Carlo sampling that allows us to simultaneously estimate the unknown continuum level of each quasar in an ensemble of high-resolution spectra, as well as their common probability distribution function (PDF) for the transmitted Lyα forest flux. This fully automated PDF regulated continuum fitting method models the unknown quasar continuum with a linear principal component analysis (PCA) basis, with the PCA coefficients treated as nuisance parameters. The method allows one to estimate parameters governing the thermal state of the intergalactic medium (IGM), such as the slope of the temperature–density relation γ−1, while marginalizing out continuum uncertainties in a fully Bayesian way. Using realistic mock quasar spectra created from a simplified semi-numerical model of the IGM, we show that this method recovers the underlying quasar continua to a precision of ≃7% and ≃10% at z = 3 and z = 5, respectively. Given the number of principal component spectra, this is comparable to the underlying accuracy of the PCA model itself. Most importantly, we show that we can achieve a nearly unbiased estimate of the slope γ−1 of the IGM temperature–density relation with a precision of ±8.6% at z = 3 and ±6.1% at z = 5, for an ensemble of ten mock high-resolution quasar spectra. Applying this method to real quasar spectra and comparing to a more realistic IGM model from hydrodynamical simulations would enable precise measurements of the thermal and cosmological parameters governing the IGM, albeit with somewhat larger uncertainties, given the increased flexibility of the model.

79 ASTRONOMY AND ASTROPHYSICS↗

Frequency Distribution of Second Solid Cancer Locations in Relation to the Irradiated Volume Among 115 Patients Treated for Childhood Cancer

Purpose: To provide better estimates of the frequency distribution of second malignant neoplasm (SMN) sites in relation to previous irradiated volumes, and better estimates of the doses delivered to these sites during radiotherapy (RT) of the first malignant neoplasm (FMN). Methods and Materials: The study focused on 115 patients who developed a solid SMN among a cohort of 4581 individuals. The homemade software package Dos{sub E}G was used to estimate the radiation doses delivered to SMN sites during RT of the FMN. Three-dimensional geometry was used to evaluate the distances between the irradiated volume, for RT delivered to each FMN, and the site of the subsequent SMN. Results: The spatial distribution of SMN relative to the irradiated volumes in our cohort was as follows: 12% in the central area of the irradiated volume, which corresponds to the planning target volume (PTV), 66% in the beam-bordering region (i.e., the area surrounding the PTV), and 22% in regions located more than 5 cm from the irradiated volume. At the SMN site, all dose levels ranging from almost zero to >75 Gy were represented. A peak SMN frequency of approximately 31% was identified in volumes that received <2.5 Gy. Conclusion: A greater volume of tissues receives low or intermediate doses in regions bordering the irradiated volume with modern multiple-beam RT arrangements. These results should be considered for risk-benefit evaluations of RT.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Efficient Levenberg-Marquardt minimization of the maximum likelihood estimator for Poisson deviates

Histograms of counted events are Poisson distributed, but are typically fitted without justification using nonlinear least squares fitting. The more appropriate maximum likelihood estimator (MLE) for Poisson distributed data is seldom used. We extend the use of the Levenberg-Marquardt algorithm commonly used for nonlinear least squares minimization for use with the MLE for Poisson distributed data. In so doing, we remove any excuse for not using this more appropriate MLE. We demonstrate the use of the algorithm and the superior performance of the MLE using simulations and experiments in the context of fluorescence lifetime imaging. Scientists commonly form histograms of counted events from their data, and extract parameters by fitting to a specified model. Assuming that the probability of occurrence for each bin is small, event counts in the histogram bins will be distributed according to the Poisson distribution. We develop here an efficient algorithm for fitting event counting histograms using the maximum likelihood estimator (MLE) for Poisson distributed data, rather than the non-linear least squares measure. This algorithm is a simple extension of the common Levenberg-Marquardt (L-M) algorithm, is simple to implement, quick and robust. Fitting using a least squares measure is most common, but it is the maximum likelihood estimator only for Gaussian-distributed data. Non-linear least squares methods may be applied to event counting histograms in cases where the number of events is very large, so that the Poisson distribution is well approximated by a Gaussian. However, it is not easy to satisfy this criterion in practice - which requires a large number of events. It has been well-known for years that least squares procedures lead to biased results when applied to Poisson-distributed data; a recent paper providing extensive characterization of these biases in exponential fitting is given. The more appropriate measure based on the maximum likelihood estimator (MLE) for the Poisson distribution is also well known, but has not become generally used. This is primarily because, in contrast to non-linear least squares fitting, there has been no quick, robust, and general fitting method. In the field of fluorescence lifetime spectroscopy and imaging, there have been some efforts to use this estimator through minimization routines such as Nelder-Mead optimization, exhaustive line searches, and Gauss-Newton minimization. Minimization based on specific one- or multi-exponential models has been used to obtain quick results, but this procedure does not allow the incorporation of the instrument response, and is not generally applicable to models found in other fields. Methods for using the MLE for Poisson-distributed data have been published by the wider spectroscopic community, including iterative minimization schemes based on Gauss-Newton minimization. The slow acceptance of these procedures for fitting event counting histograms may also be explained by the use of the ubiquitous, fast Levenberg-Marquardt (L-M) fitting procedure for fitting non-linear models using least squares fitting (simple searches obtain {approx}10000 references - this doesn't include those who use it, but don't know they are using it). The benefits of L-M include a seamless transition between Gauss-Newton minimization and downward gradient minimization through the use of a regularization parameter. This transition is desirable because Gauss-Newton methods converge quickly, but only within a limited domain of convergence; on the other hand the downward gradient methods have a much wider domain of convergence, but converge extremely slowly nearer the minimum. L-M has the advantages of both procedures: relative insensitivity to initial parameters and rapid convergence. Scientists, when wanting an answer quickly, will fit data using L-M, get an answer, and move on. Only those that are aware of the bias issues will bother to fit using the more appropriate MLE for Poisson deviates. However, since there is a simple, analytical formula for the appropriate MLE measure for Poisson deviates, it is inexcusable that least squares estimators are used almost exclusively when fitting event counting histograms. There have been ways found to use successive non-linear least squares fitting to obtain similarly unbiased results, but this procedure is justified by simulation, must be re-tested when conditions change significantly, and requires two successive fits. There is a great need for a fitting routine for the MLE estimator for Poisson deviates that has convergence domains and rates comparable to the non-linear least squares L-M fitting. We show in this report that a simple way to achieve that goal is to use the L-M fitting procedure not to minimize the least squares measure, but the MLE for Poisson deviates.

59 BASIC BIOLOGICAL SCIENCES↗

Neural Network‐Based Methods for Ocean Surface Wave Measurement Using Submarine Distributed Acoustic Sensing (DAS)

Two new data-driven models for estimating ocean surface waves from distributed acoustic sensing (DAS) submarine cable strain rate are developed using supervised machine learning on a 10-day data set collected offshore of Oliktok Point, Alaska. The new models were trained on target data from seafloor pressure moorings at three sites spaced evenly along 27.1 km of cable and were benchmarked against an empirical transfer function method previously used to estimate waves from DAS. A model which uses convolutional neural networks to transform 2-km frequency-wavenumber strain spectra to seafloor pressure spectra outperforms the benchmark in wave height prediction (RMSE of 0.15 vs. 0.41 m) and period prediction (0.29 vs. 0.37 s) when evaluated on a held-out test data set. When applied to a DAS data set collected on the same cable 2 years prior, the CNN-based model maintained similar significant wave height performance (RMSE = 0.23 m) relative to available satellite altimetry data. A two-hidden-layer, fully connected neural network which transforms 1-D strain spectra to seafloor pressure spectra also outperforms the benchmark in wave height prediction (RMSE of 0.19 vs. 0.41 m), but does not generalize as well to the prior data. Regression-based machine learning is useful for estimating waves from DAS data when the pressure-strain relationship varies temporally and spatially across different wave conditions. Models can be applied to DAS data to measure waves with higher spatial resolution and longer temporal coverage than traditional methods, which often measure waves only at a single point.

Davis, Jacob R. [Univ. of Washington, Seattle, WA ↗