Search NASA⌕ Search

SEARCH · Search NASA

Results for “Data reduction methods”

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 703 records · Page 39

A Cross-Domain Optimization Framework of PMU and Communication Placement for Multidomain Resiliency and Cost Reduction

Phasor measurement units (PMUs) play a crucial role in real-time monitoring and control of power grids. They rely on a communication network to transfer measurement data to the phasor data concentrator (PDC) for further processing and analysis. In this paper, a resilient cross-domain PMU and communication link placement method for minimizing the overall installation cost of the wide-area measurement system (WAMS) is proposed. Here, the main idea is to break down the barrier between the power grid domain and the communication domain, and consider the impact of one when design the other. The PMU placement in the power grid domain takes into account the cost of communication links by generating multiple solutions with equally minimum PMU costs for communication link placement evaluation. On the other hand, the communication link placement problem reduces the cost by customizing the routing policies based on the different roles of PMUs in grid observability. The proposed WAMS design is capable of withstanding any single component failure in the power domain (PMU failure or power branch failure) or in the communication domain (communication link failure or PDC failure). Numerical study on the IEEE 57-bus system reveals that the developed cross-domain optimization framework can significantly reduce the overall installation cost of WAMS while attaining multi-domain resiliency.

24 POWER TRANSMISSION AND DISTRIBUTION↗

A Rietveld refinement method for angular- and wavelength-dispersive neutron time-of-flight powder diffraction data

This paper introduces a two-dimensional extension of the well established Rietveld refinement method for modeling neutron time-of-flight powder diffraction data. The novel approach takes into account the variation of two parameters, diffraction angle 2θ and wavelength λ, to optimally adapt to the varying resolution function in diffraction experiments. By doing so, the refinement against angular- and wavelength-dispersive data gets rid of common data-reduction steps and also avoids the loss of high-resolution information typically introduced by integration. In a case study using a numerically simulated diffraction pattern of Rh 0.81 Fe 3.19 N taking into account the layout of the future POWTEX instrument, the profile function as parameterized in 2θ and λ is extracted. As a proof-of-concept, the resulting instrument parameterization is then utilized to perform a typical refinement of the angular- and wavelength-dispersive diffraction pattern of CuNCN, yielding excellent residuals within feasible computational efforts. Another proof-of-concept is carried out by applying the same approach to a real neutron diffraction data set of CuNCN obtained from the POWGEN instrument at the Spallation Neutron Source in Oak Ridge. The paper highlights the general importance of the novel approach for data analysis at neutron time-of-flight diffractometers and its possible inclusion within existing Rietveld software packages.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Sample preparation and biomass determination of SRF model mixture using cryogenic milling and the adapted balance method

Highlights: • An alternative sample comminution procedure for SRF is tested. • Proof of principle is shown on a SRF model mixture. • The biogenic content of the SRF is analyzed with the adapted balance method. • The novel method combines combustion analysis and a data reconciliation algorithm. • Factors for the variance of the analysis results are statistically quantified. - Abstract: The biogenic fraction of a simple solid recovered fuel (SRF) mixture (80 wt% printer paper/20 wt% high density polyethylene) is analyzed with the in-house developed adapted balance method (aBM). This fairly new approach is a combination of combustion elemental analysis (CHNS) and a data reconciliation algorithm based on successive linearisation for evaluation of the analysis results. This method shows a great potential as an alternative way to determine the biomass content in SRF. However, the employed analytical technique (CHNS elemental analysis) restricts the probed sample mass to low amounts in the range of a few hundred milligrams. This requires sample comminution to small grain sizes (<200 μm) to generate representative SRF specimen. This is not easily accomplished for certain material mixtures (e.g. SRF with rubber content) by conventional means of sample size reduction. This paper presents a proof of principle investigation of the sample preparation and analysis of an SRF model mixture with the use of cryogenic impact milling (final sample comminution) and the adapted balance method (determination of biomass content). The so derived sample preparation methodology (cutting mills and cryogenic impact milling) shows a better performance in accuracy and precision for the determination of the biomass content than one solely based on cutting mills. The results for the determination of the biogenic fraction are within 1–5% of the data obtained by the reference methods, selective dissolution method (SDM) and {sup 14}C-method ({sup 14}C-M)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Signal Extraction and Simulations for n -> p^0 y y and n -> p+p-e+e- Decays at the Jefferson Lab Eta Factory

The Jefferson Lab eta Factory (JEF) began acquiring data in early 2025. The experiment aims to give insight into the connection between Dark Matter physics models and the Standard Model by investigating rare decay processes of n and n' mesons. Several other physics motivations are also a key factor in the experiment, such as probing C and/or P violation and aspects of chiral perturbation theory. For these purposes, the forward calorimeter of the GlueX experiment in Jefferson Lab was upgraded so that it provides greater positional and energy resolution. Understanding physics-motivated cuts and background removal methods is of great importance to achieving JEF goals. Several methods have been implemented to obtain invariant mass plots for the “golden” channel of interest ¿ ¿ p0¿¿, while channels such as ¿ ¿ p+p-e+e- open a promising window into CP-violating physics. This thesis work shows a sig nificant background reduction in rare decay channels of interest, asymmetry factors comparable to recent experimental measurements, an evaluation on which analysis cuts to use after data acquisition and the likelihood of probing specific rare ¿ decays. Despite background rejection from obstructing decay channels, much remains to do to extract the p0¿¿ final-state. The asymmetry between the pion and lepton planes looks promising for p+p-e+e-; simulations show that the asymmetry is consistent with zero (no instrumental asymmetry), and the next step should include generators that model the physics of the asymmetry. This thesis work may help in the effort of probing CP-violating physics or solving the mysteries between “beyond-Standard Model” and our current understanding of physics.

Oresic, Stjepan [Univ. of Regina, SK (Canada)]↗

AIACHNE's contribution for Nuclear Energy Agency Working Party on International Nuclear Data Evaluation Co-operation Subgroup 50

The AIACHNE (AI/ML Informed cAlifornium CHi Nuclear data Experiment) project aims at designing an experiment for the 252 Cf Prompt Fission Neutron Spectrum (PFNS) that explores systematic biases in an experimental database retrieved from the EXFOR databases. To that end, machine learning (ML) methods were applied to pint-point measurement features likely related to bias. From that information, we selected a feature that should be explored by the AIACHNE experiment. Measurement features are metadata encapsulating all pertinent information about the physical measurement and analysis techniques. Examples are, for instance, what neutron and fission detectors were used for the physical metadata, and what background reduction techniques were employed for analysis techniques. Such metadata were retrieved both from EXFOR entries as well as the literature of data sets described in detail in Reference 2 (at the end of the article).

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

AIACHNE's contribution for Nuclear Energy Agency Working Party on International Nuclear Data Evaluation Co-operation Subgroup 50

The AIACHNE (AI/ML Informed cAlifornium CHi Nuclear data Experiment) project aims at designing an experiment for the 252 Cf Prompt Fission Neutron Spectrum (PFNS) that explores systematic biases in an experimental database retrieved from the EXFOR databases. To that end, machine learning (ML) methods were applied to pint-point measurement features likely related to bia. From that information, we selected a feature that should be explored by the AIACHNE experiment. Measurement features are metadata encapsulating all pertinent information about the physical measurement and analysis techniques. Examples are, for instance, what neutron and fission detectors were used for the physical metadata, and what background reduction techniques were employed for analysis techniques. Such metadata were retrieved both from EXFOR entries as well as the literature of data sets described in detail in Ref. [2]. The prerequisite for applying machine learning techniques is casting the metadata into a format that can be parsed by the algorithm. This step might seem trivial but requires to find a unique language where metadata that carry the same physics meaning across several experiments must have the same identifier. One example is, for instance, the neutron detector. As seen in Figure 1, the machine learning code identified the use of 6 Li detectors as being related to bias in some datasets of the AIACHNE 252 Cf PFNS experimental database. In fact, here are several experiments that used neutron detectors containing 6Li in the database, for instance for the example below. EXFOR format has a unique keywords describing detectors such as “SCIN” or “GLASD”. One may think that these keywords are already sufficient descriptors for ML to uniquely find an issue. However, “SCIN” (used for [3, 4]) and “GLASD” (used for [5]) fail to inform the algorithm what is the active material in the detector. And, the key common issue leading to bias in 252 Cf related to neutron detectors is not whether it is a glass detector or a scintillator. No, the issue is that 6 Li was within both detector types and that even small mistakes in the detector response functions around approximately 200 keV are amplified by the 6 Li(n,α) resonance there leading to bias in data as highlighted in Fig. 1 and Ref. [1]. Hence, the features describing the neutron detector must call out the active material in the detector, rather than the existing EXFOR detector keyword, that the ML algorithm can find physically meaningful features related to bias. The AIACHNE team used a precursor of the WPEC (Working Party on International Nuclear Data Evaluation Co-operation) SG(Subgroup)-50 format to store the metadata for the ML analysis.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Left ventricular ejection fraction determined with the simulation of a very low-dose CZT-SPECT protocol and an additional count-calibration on planar radionuclide angiographic data

Purpose: To determine whether the left ventricular ejection fractions (EFs), measured on a high-sensitivity CZT single photon emission computed tomography (SPECT)-camera with a 70% reduction in recording times and a prevention of EF overestimation through an additional count-calibration, are concordant with reference EF from planar radionuclide angiography (2D-RNA). Methods: An additional 10-minute CZT-SPECT recording was performed in patients referred to 2D-RNA for cardiomyopathy (n = 23) or chemotherapy monitoring (n = 50) with an in vivo red blood cell labeling with 850 MBq {sup 99m}TcO{sub 4}{sup −}. The EF, obtained from CZT-SPECT with 100% (SPECT100) or 30% (SPECT30) projection times and with a SPECT-count calibration on the 2D-RNA counts of corresponding cavity volumes, were compared to EF from 2D-RNA. Results: Strong and equivalent relationships were documented between the EF from 2D-RNA and the calibrated EF from SPECT100 (y = 0.89x + 6.62; R{sup 2} = 0.87) and SPECT30 (y = 0.87x + 8.40; R{sup 2} = 0.85), and the mean EF from SPECT100 (54% ± 15%) and SPECT30 (53% ± 16%) were close to that from 2D-RNA (55% ± 15%). However, upward shifts in these mean values were documented in the absence of count calibration for both SPECT100 (60% ± 18%) and SPECT30 (60% ± 18%). Conclusion: Left ventricular EF may be determined on a high-sensitivity CZT-camera, a 70% reduction in injected activities, and an additional count-calibration for further enhancing the concordance with 2D-RNA values. [French] French Abstract: Objectif: Déterminer si les fractions d'éjection ventriculaire gauche (EF) mesurées sur une caméra CZT SPECT à haute sensibilité avec une réduction de 70% des temps d'enregistrement et une prévention de la surestimation de l'EF grâce à un étalonnage supplémentaire des coups concordent avec les EF de référence obtenues par angiographie scintigraphique planaire (2D-RNA). Méthodes: Un enregistrement CZT-SPECT supplémentaire de 10 minutes a été réalisé chez les patients référés pour évaluation de cardiomyopathie (n=23) ou suivi de chimiothérapie (n = 50) par angiographie scintigraphique planaire (2D-RNA) au globules rouges marqués avec 850 MBq 99mTcO4-. Les EF, obtenues au moyen d’une caméra CZT-SPECT avec 100% (SPECT100) ou 30% (SPECT30) de temps d’acquisition et avec étalonnage des coups SPECT sur les images 2D-RNA des volumes de cavité correspondants ont été comparés aux EF à partir de 2D-RNA. Résultats: une corrélation significative existe entre les EF obtenues par 2D-RNA et les EF étalonnées obtenues par SPECT100 (y = 0,89x + 6,62; R2 = 0,87) et SPECT30 (y = 0,87x + 8,40; R2 = 0,85). Les EF moyennes de SPECT100 (54 ± 15%) et SPECT30 (53 ± 16%) sont similaires à celles obtenues par 2D-RNA (55 ± 15%). En l'absence d'étalonnage de comptage des études SPECT100 (60 ± 18%) et SPECT30 (60 ± 18%), ces valeurs moyennes d’EF sont généralement plus élevées. Conclusion: La fonction ventriculaire gauche peut être déterminée avec une caméra CZT à haute sensibilité, une réduction des activités injectées et un étalonnage supplémentaire des coups pour améliorer la concordance avec les valeurs obtenues par 2D-RNA. [Spanish] Spanish Abstract: Propósito: Determinar si la fracción de eyección del ventrículo izquierdo, medida con la cámara CZT-SPECT con una reducción del 70% en tiempo de grabación y con la prevención de la sobreestimación mediante una calibración de cuentas adicional, que sea concordante con el valor referencia de fracción de eyección mediante radiografía planar de radionucleidos. (2D-RNA). Métodos: Una grabación adicional de 10 minutos con CZT-SPECT se realizo en pacientes referidos a radiografía planar de radionucleidos (2D-RNA) por miocardiopatías. (n = 23) o para monitorizar las quimioterapias (n = 50) con marcaje de glóbulos rojos con 850 MBq 99m TcO4. La fracción de eyección obtenida por CZT-SPECT con 100% (SPECT100) o 30% (SPECT30) de tiempo de proyección, y con SPECT con calibración de cuentas con el 2D-RNA, los volúmenes de cavidades fueron comparados con fracción de eyección . Relaciones fuertes y equivalentes fueron documentados entre fracción de eyección del 2D-RNA, y la forma calibrada de la fracción de eyección SPECT100 (y = 0.89x + 6.62; R2 = 0.87) y SPECT30 y = 0.87x + 8.40; R2 = 0.85), y la media de fracción de eyección de SPECT100 (54 ± 15%) and SPECT30 (53 ± 16%) fueron cercanos a la de el 2D-RNA (55 ± 15%). Aunque, aumentos en estas medias se documentaron en la ausencia de la calibración de cuentas para SPECT100 (60 ± 18%) y SPECT30 (60 ± 18%). Conclusión: La fracción de eyección del ventrículo izquierdo puede ser determinada con una cámara CZT de alta sensibilidad, una reducción del 70% en actividades inyectadas y la calibración de cuentas adicional para realzar la concordancia con los valores de 2D-RNA. [Chinese] Chinese Abstract: 目的: 以平面放射性核素血管造影 (2D-RNA) 测定左心室射血分数 (EF) 作为参考标准,为确定利用高灵敏度 CZT SPECT 相机在减少 70% 的记录时间, 并结合计数校准防止高估情况下测量的 EF值, 是否与 2D-RNA 测得的 EF 值一致。 方法: 对心肌病 (n = 23) 或化疗监测 (n = 50) 的患者采用体内标记法注射 850 MBq {sup 99m}TcO{sub 4} 标记的红细胞进行 2D-RNA ,随后再采集 10 分钟 CZT-SPECT。 对 CZT-SPECT 100% 投影时间 (SPECT100) 或 30% 投影时间 (SPECT30) 获得的图像进行重建, 并进行计数校准 (对相应容积获得的 2D-RNA 计数通过 SPECT 校准)获得 EF 值, 与 2D-RNA 测定的 EF 的进行比较。 结果: 2D-RNA 测定的 EF 值与 SPECT100 校准后测定的 EF 值 (y = 0.89x + 6.62; R{sup 2} = 0.87) 及 SPECT30 (y = 0.87x + 8.40; R{sup 2} = 0.85) 校准后 EF 值具有高度相关性, 并且 SPECT100 (54 ± 15%) 和 SPECT30 (53 ± 16%) 测定的 EF 平均值与 2D-RNA 测定的 EF 平均值 (55 ± 15%) 非常接近。 但是, 在没有计数校准的情况下, SPECT100 (60 ± 18%) 和 SPECT30 (60 ± 18%) 测定的 EF 平均值会被高估。 结论: 高灵敏度 CZT 相机可在减少 70% 的注射剂量情况下测定左心室 EF 值, 结合计数校准可进一步提高与 2D-RNA测得的 EF 值的一致性。 关键词: 放射性核素血管造影; 血池 SPECT; CZT 相机; 左心室射血分数; 校准曲线。.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Real-Time Active Cosmic Neutron Background Reduction Methods

Neutron counting using large arrays of pressurized 3He proportional counters from an aerial system or in a maritime environment suffers from the background counts from the primary cosmic neutrons and secondary neutrons caused by cosmic ray-induced mechanisms like spallation and charge-exchange reaction. This paper reports the work performed at the Remote Sensing Laboratory–Andrews (RSL-A) and results obtained when using two different methods to reduce the cosmic neutron background in real time. Both methods used shielding materials with a high concentration (up to 30% by weight) of neutron-absorbing materials, such as natural boron, to remove the low-energy neutron flux from the cosmic background as the first step of the background reduction process. Our first method was to design, prototype, and test an up-looking plastic scintillator (BC-400, manufactured by Saint Gobain Corporation) to tag the cosmic neutrons and then create a logic pulse of a fixed time duration (~120 μs) to block the data taken by the neutron counter (pressurized 3He tubes running in a proportional counter mode). The second method examined the time correlation between the arrival of two successive neutron signals to the counting array and calculated the excess of variance (Feynman variance Y2F)1 in the neutron count distribution from Poisson distribution. The dilution of this variance from cosmic background values ideally would signal the presence of man-made neutrons.2 The first method has been technically successful in tagging the neutrons in the cosmic-ray flux and preventing them from being counted in the 3He tube array by electronic veto—field measurement work shows the efficiency of the electronic veto counter to be about 87%. The second method has successfully derived an empirical relationship between the percentile non-cosmic component in a neutron flux and the Y2F of the measured neutron count distribution. By using shielding materials alone, approximately 55% of the neutron flux from man-made sources like 252Cf or Am-Be was removed.

Neutron counting, cosmic background, ship effect, ↗

Separating Physics and Dynamics Grids for Improved Computational Efficiency in Spectral Element Earth System Models

Previous studies have shown that atmospheric models with a spectral element grid can benefit from putting physics calculations on a relatively coarse finite volume grid. Here we demonstrate an alternative high-order, element-based mapping approach used to implement a quasi-equal-area, finite volume physics grid in E3SM. Unlike similar methods, the new method in E3SM requires topology data purely local to each spectral element, which trivially allows for regional mesh refinement. Simulations with physics grids defined by 2 × 2, 3 × 3, and 4 × 4 divisions of each element are shown to verify that the alternative physics grid does not qualitatively alter the model solution. The model performance is substantially affected by the reduction of physics columns when using the 2 × 2 grid, which can increase the throughput of physics calculations by roughly 60%–120% depending on whether the computational resources are configured to maximize throughput or efficiency. A pair of regionally refined cases are also shown to highlight the refinement capability.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

MULTILEVEL ACCELERATION OF STOCHASTIC COLLOCATION METHODS FOR PDE WITH RANDOM INPUT DATA

Stochastic Collocation (SC) methods for stochastic partial differential equa- tions (SPDEs) suffer from the curse of dimensionality, whereby increases in the stochastic dimension cause an explosion of computational effort. To combat these challenges, multilevel approximation methods seek to decrease computational complexity by balancing spatial and stochastic discretization errors. As a form of variance reduction, multilevel techniques have been successfully applied to Monte Carlo (MC) methods, but may be extended to accelerate other methods for SPDEs in which the stochastic and spatial degrees of freedom are de- coupled. This article presents general convergence and computational complexity analysis of a multilevel method for SPDEs, demonstrating its advantages with regard to standard, single level approximation. The numerical results will highlight conditions under which multilevel sparse grid SC is preferable to the more traditional MC and SC approaches.

Webster, Clayton G↗

SU-E-T-395: Multi-GPU-Based VMAT Treatment Plan Optimization Using a Column-Generation Approach

Purpose: GPU has been employed to speed up VMAT optimizations from hours to minutes. However, its limited memory capacity makes it difficult to handle cases with a huge dose-deposition-coefficient (DDC) matrix, e.g. those with a large target size, multiple arcs, small beam angle intervals and/or small beamlet size. We propose multi-GPU-based VMAT optimization to solve this memory issue to make GPU-based VMAT more practical for clinical use. Methods: Our column-generation-based method generates apertures sequentially by iteratively searching for an optimal feasible aperture (referred as pricing problem, PP) and optimizing aperture intensities (referred as master problem, MP). The PP requires access to the large DDC matrix, which is implemented on a multi-GPU system. Each GPU stores a DDC sub-matrix corresponding to one fraction of beam angles and is only responsible for calculation related to those angles. Broadcast and parallel reduction schemes are adopted for inter-GPU data transfer. MP is a relatively small-scale problem and is implemented on one GPU. One headand- neck cancer case was used for test. Three different strategies for VMAT optimization on single GPU were also implemented for comparison: (S1) truncating DDC matrix to ignore its small value entries for optimization; (S2) transferring DDC matrix part by part to GPU during optimizations whenever needed; (S3) moving DDC matrix related calculation onto CPU. Results: Our multi-GPU-based implementation reaches a good plan within 1 minute. Although S1 was 10 seconds faster than our method, the obtained plan quality is worse. Both S2 and S3 handle the full DDC matrix and hence yield the same plan as in our method. However, the computation time is longer, namely 4 minutes and 30 minutes, respectively. Conclusion: Our multi-GPU-based VMAT optimization can effectively solve the limited memory issue with good plan quality and high efficiency, making GPUbased ultra-fast VMAT planning practical for real clinical use.

60 APPLIED LIFE SCIENCES↗

Progressive cone beam CT dose control in image-guided radiation therapy

Purpose: Cone beam CT (CBCT) in image-guided radiotherapy (IGRT) offers a tremendous advantage for treatment guidance. The associated imaging dose is a clinical concern. One unique feature of CBCT-based IGRT is that the same patient is repeatedly scanned during a treatment course, and the contents of CBCT images at different fractions are similar. The authors propose a progressive dose control (PDC) scheme to utilize this temporal correlation for imaging dose reduction. Methods: A dynamic CBCT scan protocol, as opposed to the static one in the current clinical practice, is proposed to gradually reduce the imaging dose in each treatment fraction. The CBCT image from each fraction is processed by a prior-image based nonlocal means (PINLM) module to enhance its quality. The increasing amount of prior information from previous CBCT images prevents degradation of image quality due to the reduced imaging dose. Two proof-of-principle experiments have been conducted using measured phantom data and Monte Carlo simulated patient data with deformation. Results: In the measured phantom case, utilizing a prior image acquired at 0.4 mAs, PINLM is able to improve the image quality of a CBCT acquired at 0.2 mAs by reducing the noise level from 34.95 to 12.45 HU. In the synthetic patient case, acceptable image quality is maintained at four consecutive fractions with gradually decreasing exposure levels of 0.4, 0.1, 0.07, and 0.05 mAs. When compared with the standard low-dose protocol of 0.4 mAs for each fraction, an overall imaging dose reduction of more than 60% is achieved. Conclusions: PINLM-PDC is able to reduce CBCT imaging dose in IGRT utilizing the temporal correlations among the sequence of CBCT images while maintaining the quality.

61 RADIATION PROTECTION AND DOSIMETRY↗

Ionization potential and electron affinity for six common explosive compounds by DFT, MP2, and CBS-QB3

The vertical and adiabatic ionization potential (IP V and IP A ) and vertical electron affinity (EA V ) for six explosives (RDX, HMX, TNT, PETN, HMTD, and TATP) have been studied by ab initio computational methods. The IPV was calculated using MP2 and CBS-QB3 while the IP A was calculated with B3LYP, CAM-B3LYP, ω B97XD, B2PLYP, and MP2. RDX and TNT IP A ’s were also reported using CBS -QB3. Excluding results by CBS-QB3, B3LYP and B2PLYPD provided superior and more consistent results for calculating the IP. The EA V were calculated using the same methods however B3LYP performed the worst in this case with MP2 and B2PLYPD predicting values closest to those made by CBS-QB3, which was used a reference due to lacking experimental data. Basis set effects were evaluated using 6- 31+G(d,p), 6-311+G(d,p), and 6-311+(3df,2p) for both IP and EA. 6-31+G(d,p) gave satisfactory results for calculating both IP however 6-311+G(3df,2p) had improved results in calculating the EA. The four nitro containing compounds had exothermic reduction potentials while the peroxides were unfavorable. Additionally, it was seen that RDX, HMX, TATP and HMTD were unstable in their reduced forms. Results are aimed to assist detection and screening methods.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Monotonic Gaussian Process for Physics-Constrained Machine Learning With Materials Science Applications

Physics-constrained machine learning is emerging as an important topic in the field of machine learning for physics. One of the most significant advantages of incorporating physics constraints into machine learning methods is that the resulting model requires significantly less data to train. By incorporating physical rules into the machine learning formulation itself, the predictions are expected to be physically plausible. Gaussian process (GP) is perhaps one of the most common methods in machine learning for small datasets. In this paper, we investigate the possibility of constraining a GP formulation with monotonicity on three different material datasets, where one experimental and two computational datasets are used. The monotonic GP is compared against the regular GP, where a significant reduction in the posterior variance is observed. The monotonic GP is strictly monotonic in the interpolation regime, but in the extrapolation regime, the monotonic effect starts fading away as one goes beyond the training dataset. Imposing monotonicity on the GP comes at a small accuracy cost, compared to the regular GP. The monotonic GP is perhaps most useful in applications where data are scarce and noisy, and monotonicity is supported by strong physical evidence.

36 MATERIALS SCIENCE↗

Reconstruction of jointly sparse vectors via manifold optimization

In this paper, we consider the challenge of reconstructing jointly sparse vectors from linear measurements. Firstly, we show that by utilizing the rank of the output data matrix we can reduce the problem to a full column rank case. This result reveals a reduction in the computational complexity of the original problem and enables a simple implementation of joint sparse recovery algorithms for full-rank setting. Secondly, we propose a new method for joint sparse recovery in the form of a non-convex optimization problem on a non-compact Stiefel manifold. In our numerical experiments our method outperforms the commonly used l 2,1 minimization in the sense that much fewer measurements are required for accurate sparse reconstructions. Here, we postulate this approach possesses the desirable rank aware property, that is, being able to take advantage of the rank of the unknown matrix to improve the recovery.

97 MATHEMATICS AND COMPUTING↗

NEW APPROACHES TO PHOTOMETRIC REDSHIFT PREDICTION VIA GAUSSIAN PROCESS REGRESSION IN THE SLOAN DIGITAL SKY SURVEY

Expanding upon the work of Way and Srivastava we demonstrate how the use of training sets of comparable size continue to make Gaussian process regression (GPR) a competitive approach to that of neural networks and other least-squares fitting methods. This is possible via new large-size matrix inversion techniques developed for Gaussian processes (GPs) that do not require that the kernel matrix be sparse. This development, combined with a neural-network kernel function appears to give superior results for this problem. Our best-fit results for the Sloan Digital Sky Survey (SDSS) Main Galaxy Sample using u, g, r, i, z filters gives an rms error of 0.0201 while our results for the same filters in the luminous red galaxy sample yield 0.0220. We also demonstrate that there appears to be a minimum number of training-set galaxies needed to obtain the optimal fit when using our GPR rank-reduction methods. We find that morphological information included with many photometric surveys appears, for the most part, to make the photometric redshift evaluation slightly worse rather than better. This would indicate that most morphological information simply adds noise from the GP point of view in the data used herein. In addition, we show that cross-match catalog results involving combinations of the Two Micron All Sky Survey, SDSS, and Galaxy Evolution Explorer have to be evaluated in the context of the resulting cross-match magnitude and redshift distribution. Otherwise one may be misled into overly optimistic conclusions.

79 ASTRONOMY AND ASTROPHYSICS↗

Analysis and Benchmarking of feature reduction for classification under computational constraints

Abstract Machine learning is most often expensive in terms of computational and memory costs due to training with large volumes of data. Current computational limitations of many computing systems motivate us to investigate practical approaches, such as feature selection and reduction, to reduce the time and memory costs while not sacrificing the accuracy of classification algorithms. In this work, we carefully review, analyze, and identify the feature reduction methods that have low costs/overheads in terms of time and memory. Then, we evaluate the identified reduction methods in terms of their impact on the accuracy, precision, time, and memory costs of traditional classification algorithms. Specifically, we focus on the least resource intensive feature reduction methods that are available in Scikit-Learn library. Since our goal is to identify the best performing low-cost reduction methods, we do not consider complex expensive reduction algorithms in this study. In our evaluation, we find that at quadratic-scale feature reduction, the classification algorithms achieve the best trade-off among competitive performance metrics. Results show that the overall training times are reduced 61%, the model sizes are reduced 6×, and accuracy scores increase 25% compared to the baselines on average with quadratic scale reduction.

97 MATHEMATICS AND COMPUTING↗

Efficient Uncertainty Quantification in Stochastic Economic Dispatch

Stochastic economic dispatch models usually address uncertainties in forecasts of renewable generation output by considering a finite number of realizations drawn from a stochastic process model, typically via Monte Carlo sampling. Accurate evaluations of expectations or higher order moments for quantities of interest, e.g., generating cost, can require a prohibitively large number of samples. We propose an alternative to Monte Carlo sampling based on polynomial chaos expansions. These representations enable efficient and accurate propagation of uncertainties in model parameters, using sparse quadrature methods. Furthermore, we use Karhunen-Loève expansions for efficient representation of uncertain renewable energy generation that follows geographical and temporal correlations derived from historical data at each wind farm. Considering expected production cost, we show that the proposed approach can yield several orders of magnitude reduction in computational cost for solving stochastic economic dispatch relative to Monte Carlo sampling, for a given target error threshold.

24 POWER TRANSMISSION AND DISTRIBUTION↗