Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Likelihood Functions”

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 271 records · Page 15

Shape restricted additive hazards models: Monotone, unimodal, and U‐shape hazard functions

We consider estimation of the semiparametric additive hazards model with an unspecified baseline hazard function where the effect of a continuous covariate has a specific shape but otherwise unspecified. Such estimation is particularly useful for a unimodal hazard function, where the hazard is monotone increasing and monotone decreasing with an unknown mode. A popular approach of the proportional hazards model is limited in such setting due to the complicated structure of the partial likelihood. Our model defines a quadratic loss function, and its simple structure allows a global Hessian matrix that does not involve parameters. Thus, once the global Hessian matrix is computed, a standard quadratic programming method can be applicable by profiling all possible locations of the mode. However, the quadratic programming method may be inefficient to handle a large global Hessian matrix in the profiling algorithm due to a large dimensionality, where the dimension of the global Hessian matrix and number of hypothetical modes are the same order as the sample size. We propose the quadratic pool adjacent violators algorithm to reduce computational costs. The proposed algorithm is extended to the model with a time‐dependent covariate with monotone or U‐shape hazard function. In simulation studies, our proposed method improves computational speed compared to the quadratic programming method, with bias and mean square error reductions. We analyze data from a recent cardiovascular study.

Mathematical & Computational Biology↗

A selection function toolbox for subsets of astronomical catalogues

ABSTRACT Large catalogues are ubiquitous throughout astronomy, but most scientific analyses are carried out on smaller samples selected from these catalogues by chosen cuts on catalogued quantities. The selection function of that scientific sample – the probability that a star in the catalogue will satisfy these cuts and so make it into the sample – is thus unique to each scientific analysis. We have created a general framework that can flexibly estimate the selection function of a sample drawn from a catalogue in terms of position, magnitude and colour. Our method is unique in using the binomial likelihood and accounting for correlations in the selection function across position, magnitude, and colour using Gaussian processes and spherical harmonics. We have created a new open-source python package selectionfunctiontoolbox that implements this framework and used it to make three different estimates of the APOGEE DR16 red giant sample selection function, as a subset of 2MASS, with each estimate using an increasing amount of technical knowledge of the APOGEE targeting. In a companion paper, we applied our methodology to derive estimates of the astrometric and spectroscopic selection functions of Gaia EDR3. Our framework will make it trivial for astrophysicists to estimate the selection function that they should be using with the custom sample of stars that they have chosen to answer their scientific question.

79 ASTRONOMY AND ASTROPHYSICS↗

How are coastal households responding to climate change?

In Australia, shared responsibility is a concept advocated to promote collective climate change adaptation by multiple actors and institutions. However, a shared response is often promoted in the absence of information regarding actions currently taken; in particular, there is limited knowledge regarding action occurring at the household scale. To address this gap, we examine household actions taken to address climate change and associated hazards in two Australian coastal communities. Mixed methods research is conducted to answer three questions: (1) what actions are currently taken (mitigation, actions to lobby for change or adaptation to climate impacts)? (2) why are these actions taken (e.g. are they consistent with capacity, experience, perceptions of risk); and (3) what are the implications for adaptation? We find that households are predominantly mitigating greenhouse gas emissions and that impact orientated adaptive actions are limited. Coping strategies are considered sufficient to mange climate risks, proving a disincentive for additional adaptive action. Influencing factors differ, but generally, risk perception and climate change belief are associated with action. Furthermore, the likelihood of more action is a function of homeownership and a tendency to plan ahead. Addressing factors that support or constrain household adaptive decision-making and action, from the physical (e.g. homeownership) to the social (e.g. skills in planning and a culture of adapting to change) will be critical in increasing household participation in adaptation.

54 ENVIRONMENTAL SCIENCES↗

Magnetic character of the low-energy enhancement in 70 Zn

The transition of an atomic nucleus from an excited state to a state of lower energy is typically accompanied by the emission of photons. The likelihood of photon emission is a function of the photon energy and of properties of the initial and final nuclear states and is known as the photon (or γ-ray) strength function (γSF). Of the features present in the γSF, an enhancement in the low-energy region has been observed in some nuclei, but the electromagnetic nature of this enhancement remains an open question. This low-energy enhancement in the γSF significantly impacts our understanding of how elements are created in stars and the structure of the nucleus. Here we present the results from an experiment on the γSF of 70 Zn that show that the enhancement in the γSF of this nucleus is of magnetic nature. This observation helps address a long-open question about γSFs in atomic nuclei and directly impacts our predictive capabilities in nuclear science.

Physics - Nuclear physics and radiation physics↗

HST/WFC3 OBSERVATIONS OF LOW-MASS GLOBULAR CLUSTERS AM 4 AND PALOMAR 13: PHYSICAL PROPERTIES AND IMPLICATIONS FOR MASS LOSS

We investigate the loss of low-mass stars in two of the faintest globular clusters known, AM 4 and Palomar 13 (Pal 13), using HST/WFC3 F606W and F814W photometry. To determine the physical properties of each cluster—age, mass, metallicity, extinction, and present day mass function (MF)—we use the maximum likelihood color-magnitude diagram (CMD) fitting program MATCH and the Dartmouth, Padova, and BaSTI stellar evolution models. For AM 4, the Dartmouth models provide the best match to the CMD and yield an age of >13 Gyr, metallicity log Z/Z {sub ☉} = –1.68 ± 0.08, a distance modulus (m – M) {sub V} = 17.47 ± 0.03, and reddening A{sub V} = 0.19 ± 0.02. For Pal 13 the Dartmouth models give an age of 13.4 ± 0.5 Gyr, log Z/Z {sub ☉} = –1.55 ± 0.06, (m – M) {sub V} = 17.17 ± 0.02, and A{sub V} = 0.43 ± 0.01. We find that the systematic uncertainties due to choice in assumed stellar model greatly exceed the random uncertainties, highlighting the importance of using multiple stellar models when analyzing stellar populations. Assuming a single-sloped power-law MF, we find that AM 4 and Pal 13 have spectral indices α = +0.68 ± 0.34 and α = –1.67 ± 0.25 (where a Salpeter MF has α = +1.35), respectively. Comparing our derived slopes with literature measurements of cluster integrated magnitude (M{sub V} ) and MF slope indicates that AM 4 is an outlier. Its MF slope is substantially steeper than clusters of comparable luminosity, while Pal 13 has an MF in line with the general trend. We discuss both primordial and dynamical origins for the unusual MF slope of AM 4 and tentatively favor the dynamical scenario. However, MF slopes of more low luminosity clusters are needed to verify this hypothesis.

79 ASTRONOMY AND ASTROPHYSICS↗

LOCAL TWO-DIMENSIONAL PARTICLE-IN-CELL SIMULATIONS OF THE COLLISIONLESS MAGNETOROTATIONAL INSTABILITY

The magnetorotational instability (MRI) is a crucial mechanism of angular momentum transport in a variety of astrophysical accretion disks. In systems accreting at well below the Eddington rate, such as the central black hole in the Milky Way (Sgr A*), the plasma in the disk is essentially collisionless. We present a nonlinear study of the collisionless MRI using first-principles particle-in-cell plasma simulations. We focus on local two-dimensional (axisymmetric) simulations, deferring more realistic three-dimensional simulations to future work. For simulations with net vertical magnetic flux, the MRI continuously amplifies the magnetic field, B, until the Alfven velocity, v{sub A} , is comparable to the speed of light, c (independent of the initial value of v{sub A} /c). This is consistent with the lack of saturation of MRI channel modes in analogous axisymmetric MHD simulations. The amplification of the magnetic field by the MRI generates a significant pressure anisotropy in the plasma (with the pressure perpendicular to B being larger than the parallel pressure). We find that this pressure anisotropy in turn excites mirror modes and that the volume-averaged pressure anisotropy remains near the threshold for mirror mode excitation. Particle energization is due to both reconnection and viscous heating associated with the pressure anisotropy. Reconnection produces a distinctive power-law component in the energy distribution function of the particles, indicating the likelihood of non-thermal ion and electron acceleration in collisionless accretion disks. This has important implications for interpreting the observed emission-from the radio to the gamma-rays-of systems such as Sgr A*.

79 ASTRONOMY AND ASTROPHYSICS↗

A new perspective on dark energy modeling via genetic algorithms

We use Genetic Algorithms to extract information from several cosmological probes, such as the type Ia supernovae (SnIa), the Baryon Acoustic Oscillations (BAO) and the growth rate of matter perturbations. This is done by implementing a model independent and bias-free reconstruction of the various scales and distances that characterize the data, like the luminosity d{sub L}(z) and the angular diameter distance d{sub A}(z) in the SnIa and BAO data, respectively, or the dependence with redshift of the matter density Ω{sub m}(a) in the growth rate data, fσ{sub 8}(z). These quantities can then be used to reconstruct the expansion history of the Universe, and the resulting Dark Energy (DE) equation of state w(z) in the context of FRW models, or the mass radial function Ω{sub M}(r) in LTB models. In this way, the reconstruction is completely independent of our prior bias. Furthermore, we use this method to test the Etherington relation, ie the well-known relation between the luminosity and the angular diameter distance, η≡d{sub L}(z)/(1+z){sup 2}d{sub A}(z), which is equal to 1 in metric theories of gravity. We find that the present data seem to suggest a 3-σ deviation from one at redshifts z ∼ 0.5. Finally, we present a novel way, within the Genetic Algorithm paradigm, to analytically estimate the errors on the reconstructed quantities by calculating a Path Integral over all possible functions that may contribute to the likelihood. We show that this can be done regardless of the data being correlated or uncorrelated with each other and we also explicitly demonstrate that our approach is in good agreement with other error estimation techniques like the Fisher Matrix approach and the Bootstrap Monte Carlo.

79 ASTRONOMY AND ASTROPHYSICS↗

Analysis of Polarimetry Data with Angular Uncertainties

For a track based polarimeter, such as the Imaging X-ray Polarimetry Explorer (IXPE), the sensitivity to polarization depends on the modulation factor, which is a strong function of energy. In previous work, a likelihood method was developed that would account for this variation in order to estimate the minimum detectable polarization (MDP). That method essentially required that the position angles of individual events should be known precisely. In a separate work, however, it was shown that using a machine-learning method for measuring event tracks can generate track angle uncertainties, which can be used in the analysis. Here, the maximum likelihood method is used as a basis for revising the estimate of the MDP in a general way that can include uncertainties in event track position angles. The resultant MDP depends solely upon the distribution of track angle uncertainties present in the input data. Due to the physics of the IXPE detectors, it is possible to derive a simple relationship between these angular uncertainties and the energy-dependent modulation function as a step in the process.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Bayesian inference of nonlinear unsteady aerodynamics from aeroelastic limit cycle oscillations

A Bayesian model selection and parameter estimation algorithm is applied to investigate the influence of nonlinear and unsteady aerodynamic loads on the limit cycle oscillation (LCO) of a pitching airfoil in the transitional Reynolds number regime. At small angles of attack, laminar boundary layer trailing edge separation causes negative aerodynamic damping leading to the LCO. The fluid–structure interaction of the rigid, but elastically mounted, airfoil and nonlinear unsteady aerodynamics is represented by two coupled nonlinear stochastic ordinary differential equations containing uncertain parameters and model approximation errors. Several plausible aerodynamic models with increasing complexity are proposed to describe the aeroelastic system leading to LCO. The likelihood in the posterior parameter probability density function (pdf) is available semi-analytically using the extended Kalman filter for the state estimation of the coupled nonlinear structural and unsteady aerodynamic model. The posterior parameter pdf is sampled using a parallel and adaptive Markov Chain Monte Carlo (MCMC) algorithm. The posterior probability of each model is estimated using the Chib–Jeliazkov method that directly uses the posterior MCMC samples for evidence (marginal likelihood) computation. The Bayesian algorithm is validated through a numerical study and then applied to model the nonlinear unsteady aerodynamic loads using wind-tunnel test data at various Reynolds numbers.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

THE CANADA-FRANCE HIGH-z QUASAR SURVEY: NINE NEW QUASARS AND THE LUMINOSITY FUNCTION AT REDSHIFT 6

We present discovery imaging and spectroscopy for nine new z {approx} 6 quasars found in the Canada-France High-z Quasar Survey (CFHQS) bringing the total number of CFHQS quasars to 19. By combining the CFHQS with the more luminous Sloan Digital Sky Survey sample, we are able to derive the quasar luminosity function from a sample of 40 quasars at redshifts 5.74 < z < 6.42. Our binned luminosity function shows a slightly lower normalization and flatter slope than found in previous work. The binned data also suggest a break in the luminosity function at M {sub 1450} {approx} -25. A double power-law maximum likelihood fit to the data is consistent with the binned results. The luminosity function is strongly constrained (1{sigma} uncertainty <0.1 dex) over the range -27.5 < M {sub 1450} < -24.7. The best-fit parameters are {phi}(M*{sub 1450}) = 1.14 x 10{sup -8} Mpc{sup -3} mag{sup -1}, break magnitude M*{sub 1450} = -25.13, and bright end slope {beta} = -2.81. However, the covariance between {beta} and M*{sub 1450} prevents strong constraints being placed on either parameter. For a break magnitude in the range -26 < M*{sub 1450} < -24, we find -3.8 < {beta} < -2.3 at 95% confidence. We calculate the z = 6 quasar intergalactic ionizing flux and show it is between 20 and 100 times lower than that necessary for reionization. Finally, we use the luminosity function to predict how many higher redshift quasars may be discovered in future near-IR imaging surveys.

79 ASTRONOMY AND ASTROPHYSICS↗

THE EXOPLANET CENSUS: A GENERAL METHOD APPLIED TO KEPLER

We develop a general method to fit the underlying planetary distribution function (PLDF) to exoplanet survey data. This maximum likelihood method accommodates more than one planet per star and any number of planet or target star properties. We apply the method to announced Kepler planet candidates that transit solar-type stars. The Kepler team's estimates of the detection efficiency are used and are shown to agree with theoretical predictions for an ideal transit survey. The PLDF is fit to a joint power law in planet radius, down to 0.5 R{sub Circled-Plus }, and orbital period, up to 50 days. The estimated number of planets per star in this sample is {approx}0.7-1.4, where the range covers systematic uncertainties in the detection efficiency. To analyze trends in the PLDF we consider four planet samples, divided between shorter and longer periods at 7 days and between large and small radii at 3 R{sub Circled-Plus }. The size distribution changes appreciably between these four samples, revealing a relative deficit of {approx}3 R{sub Circled-Plus} planets at the shortest periods. This deficit is suggestive of preferential evaporation and sublimation of Neptune- and Saturn-like planets. If the trend and explanation hold, it would be spectacular observational support of the core accretion and migration hypotheses, and would allow refinement of these theories.

79 ASTRONOMY AND ASTROPHYSICS↗

Predicting the redshift 2 Hα luminosity function using [O iii] emission line galaxies

Upcoming space-based surveys such as Euclid and WFIRST-AFTA plan to measure baryonic acoustic oscillations in order to study dark energy. These surveys will use IR slitless grism spectroscopy to measure redshifts of a large number of galaxies over a significant redshift range. In this paper, we use the Wide Field Camera 3 Infrared Spectroscopic Parallel Survey (WISP) to estimate the expected number of Hα emitters observable by these future surveys. WISP is an ongoing Hubble Space Telescope slitless spectroscopic survey, covering the 0.8–1.65 μm wavelength range and allowing the detection of Hα emitters up to z ∼ 1.5 and [O iii] emitters to z ∼ 2.3. We derive the Hα–[O iii] bivariate line luminosity function (LLF) for WISP galaxies at z ∼ 1 using a maximum likelihood estimator that properly accounts for uncertainties in line luminosity measurements and we demonstrate how it can be used to derive the Hα luminosity function by exclusively fitting [O iii] data. Using the z∼2 [O iii] LLF and assuming that the relation between Hα and [O iii] luminosity does not change significantly over the redshift range, we predict the Hα number counts at z∼2—the upper end of the redshift range of interest for future surveys. For the redshift range 0.7<z<2, we expect ∼3000 galaxies deg{sup −2} for a flux limit of 3 × 10{sup −16} erg s{sup −1} cm{sup −2} (the proposed depth of the Euclid galaxy redshift survey) and ∼20,000 galaxies deg{sup −2} for a flux limit of ∼10{sup −16} erg s{sup −1} cm{sup −2} (the baseline depth of the WFIRST galaxy redshift survey).

79 ASTRONOMY AND ASTROPHYSICS↗

Learning likelihood ratios with neural network classifiers

The likelihood ratio is a crucial quantity for statistical inference in science that enables hypothesis testing, construction of confidence intervals, reweighting of distributions, and more. Many modern scientific applications, however, make use of data- or simulation-driven models for which computing the likelihood ratio can be very difficult or even impossible. By applying the so-called “likelihood ratio trick,” approximations of the likelihood ratio may be computed using clever parametrizations of neural network-based classifiers. A number of different neural network setups can be defined to satisfy this procedure, each with varying performance in approximating the likelihood ratio when using finite training data. We present a series of empirical studies detailing the performance of several common loss functionals and parametrizations of the classifier output in approximating the likelihood ratio of two univariate and multivariate Gaussian distributions as well as simulated high-energy particle physics datasets.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

A COMPARATIVE STUDY OF OPTICAL/ULTRAVIOLET VARIABILITY OF NARROW-LINE SEYFERT 1 AND BROAD-LINE SEYFERT 1 ACTIVE GALACTIC NUCLEI

The ensemble optical/ultraviolet (UV) variability of narrow-line Seyfert 1 (NLS1)-type active galactic nuclei (AGNs) is investigated, based on a sample selected from the Sloan Digital Sky Survey (SDSS) Stripe 82 region with multi-epoch photometric scanning data. As a comparison, a control sample of broad-line Seyfert 1 (BLS1)-type AGNs is also incorporated. To quantify properly the intrinsic variation amplitudes and their uncertainties, a novel method of parametric maximum likelihood is introduced that has, as we argued, certain virtues over previously used methods. The majority of NLS1-type AGNs exhibit significant variability on timescales from about 10 days to a few years with, however, smaller amplitudes on average compared to BLS1-type AGNs. About 20 NLS1-type AGNs that show relatively large variations are presented and may deserve future monitoring observations, for instance, reverberation mapping. The averaged structure functions of variability, constructed using the same maximum likelihood method, show remarkable similarity in shape for the two types of AGNs on timescales longer than about 10 days, which can be approximated by a power law or an exponential function. This, along with other similar properties, such as the wavelength-dependent variability, is indicative of a common dominant mechanism responsible for the long-term optical/UV variability of both NLS1- and BLS1-type AGNs. Toward the short timescales, however, there is tentative evidence that the structure function of NLS1-type AGNs continues to decline, whereas that of BLS1-type AGNs flattens with some residual variability on timescales of days. If this can be confirmed, it may suggest that an alternative mechanism, such as X-ray reprocessing, starts to dominate in BLS1-type AGNs, but not in NLS1-type AGNs, on such timescales.

79 ASTRONOMY AND ASTROPHYSICS↗

THE PROPERTIES OF X-RAY COLD FRONTS IN A STATISTICAL SAMPLE OF SIMULATED GALAXY CLUSTERS

We examine the incidence of cold fronts in a large sample of galaxy clusters extracted from a (512 h {sup -1} Mpc) hydrodynamic/N-body cosmological simulation with adiabatic gas physics computed with the Enzo adaptive mesh refinement code. This simulation contains a sample of roughly 4000 galaxy clusters with M {>=}10{sup 14} M{sub sun} at z = 0. For each simulated galaxy cluster, we have created mock 0.3-8.0 keV X-ray observations and spectroscopic-like temperature maps. We have searched these maps with a new automated algorithm to identify the presence of cold fronts in projection. Using a threshold of a minimum of 10 cold front pixels in our images, corresponding to a total comoving length L{sub cf}>156 h {sup -1} kpc, we find that roughly 10%-12% of all projections in a mass-limited sample would be classified as cold front clusters. Interestingly, the fraction of clusters with extended cold front features in our synthetic maps of a mass-limited sample trends only weakly with redshift out to z = 1.0. However, when using different selection functions, including a simulated flux limit, the trending with redshift changes significantly. The likelihood of finding cold fronts in the simulated clusters in our sample is a strong function of cluster mass. In clusters with M>7.5 x 10{sup 14} M{sub sun} the cold front fraction is 40%-50%. We also show that the presence of cold fronts is strongly correlated with disturbed morphology as measured by quantitative structure measures. Finally, we find that the incidence of cold fronts in the simulated cluster images is strongly dependent on baryonic physics.

79 ASTRONOMY AND ASTROPHYSICS↗

Solving inverse problems in stochastic models using deep neural networks and adversarial training

Inverse problems associated with stochastic models constitute a significant portion of scientific and engineering applications. In such cases the unknown quantities are distributions. The applicability of traditional methods is limited because of their demanding assumptions or prohibitive computational consumption; for example, maximum likelihood methods require closed-form density functions, and Markov Chain Monte Carlo needs a large number of simulations. We propose a new method that estimates the unknown distribution by matching the statistical properties between observed and simulated random processes. We leverage the expressive power of neural networks to approximate the unknown distribution and use a discriminative neural network for computing the statistical discrepancies between the observed and simulated random processes. Here we demonstrated numerically that the proposed methods can estimate both the model parameters and learn complicated unknown distributions.

42 ENGINEERING↗

Latent map Gaussian processes for mixed variable metamodeling

Gaussian processes (GPs) are ubiquitously used in sciences and engineering as metamodels. Standard GPs, however, can only handle numerical or quantitative variables. Here we introduce latent map Gaussian processes (LMGPs) that inherit the attractive properties of GPs and are also applicable to mixed data which have both quantitative and qualitative inputs. The core idea behind LMGPs is to learn a continuous, low-dimensional latent space or manifold which encodes all qualitative inputs. To learn this manifold, we first assign a unique prior vector representation to each combination of qualitative inputs. We then use a low-rank linear map to project these priors on a manifold that characterizes the posterior representations. As the posteriors are quantitative, they can be directly used in any standard correlation function such as the Gaussian or Matern. Hence, the optimal map and the corresponding manifold, along with other hyperparameters of the correlation function, can be systematically learned via maximum likelihood estimation. Through a wide range of analytic and real-world examples, we demonstrate the advantages of LMGPs over state-of-the-art methods in terms of accuracy and versatility. In particular, we show that LMGPs can handle variable-length inputs, have an explainable neural network interpretation, and provide insights into how qualitative inputs affect the response or interact with each other. We also employ LMGPs in Bayesian optimization and illustrate that they can discover optimal compound compositions more efficiently than conventional methods that convert compositions to qualitative variables via manual featurization.

42 ENGINEERING↗