Search NASA⌕ Search

SEARCH · Search NASA

Results for “model interpretability”

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 1,675 records · Page 93

DEEP CHANDRA OBSERVATIONS OF THE CRAB-LIKE PULSAR WIND NEBULA G54.1+0.3 AND SPITZER SPECTROSCOPY OF THE ASSOCIATED INFRARED SHELL

G54.1+0.3 is a young pulsar wind nebula (PWN), closely resembling the Crab, for which no thermal shell emission has been detected in X-rays. Recent Spitzer observations revealed an infrared (IR) shell containing a dozen point sources arranged in a ring-like structure, previously proposed to be young stellar objects. An extended knot of emission located in the NW part of the shell appears to be aligned with the pulsar's X-ray jet, suggesting a possible interaction with the shell material. Surprisingly, the IR spectrum of the knot resembles the spectrum of freshly formed dust in Cas A, and is dominated by an unidentified dust emission feature at 21 {mu}m. The spectra of the shell also contain various emission lines and show that some are significantly broadened, suggesting that they originate in rapidly expanding supernova (SN) ejecta. We present the first evidence that the PWN is driving shocks into expanding SN ejecta and we propose an alternative explanation for the origin of the IR emission in which the shell is composed entirely of SN ejecta. In this scenario, the freshly formed SN dust is being heated by early-type stars belonging to a cluster in which the SN exploded. Simple dust models show that this interpretation can give rise to the observed shell emission and the IR point sources.

79 ASTRONOMY AND ASTROPHYSICS↗

THE LATE PEAKING AFTERGLOW OF GRB 100418A

GRB 100418A is a long gamma-ray burst (GRB) at redshift z = 0.6235 discovered with the Swift Gamma-ray Burst Explorer with unusual optical and X-ray light curves. After an initial short-lived, rapid decline in X-rays, the optical and X-ray light curves observed with Swift are approximately flat or rising slightly out to at least {approx}7 x 10{sup 3} s after the trigger, peak at {approx}5 x 10{sup 4} s, and then follow an approximately power-law decay. Such a long optical plateau and late peaking is rarely seen in GRB afterglows. Observations with Rapid Eye Mount during a gap in the Swift coverage indicate a bright optical flare at {approx}2.5 x 10{sup 4} s. The long plateau phase of the afterglow is interpreted using either a model with continuous injection of energy into the forward shock of the burst or a model in which the jet of the burst is viewed off-axis. In both models the isotropic kinetic energy in the late afterglow after the plateau phase is {>=}10{sup 2} times the 10{sup 51} erg of the prompt isotropic gamma-ray energy release. The energy injection model is favored because the off-axis jet model would require the intrinsic T{sub 90} for the GRB jet viewed on-axis to be very short, {approx}10 ms, and the intrinsic isotropic gamma-ray energy release and the true jet energy to be much higher than the typical values of known short GRBs. The non-detection of a jet break up to t {approx} 2 x 10{sup 6} s indicates a jet half-opening angle of at least {approx}14{sup 0}, and a relatively high-collimation-corrected jet energy of E{sub jet} {>=} 10{sup 52} erg.

79 ASTRONOMY AND ASTROPHYSICS↗

AN UPSCATTERING SPECTRAL FORMATION MODEL FOR THE PROMPT EMISSION OF GAMMA-RAY BURSTS

We propose a model for the spectral formation of gamma-ray burst (GRB) prompt emission, where the phenomenological Band function is usually applied to describe this emission. We suggest that the GRB prompt emission is mainly a result of two upscattering processes. The first process is the Comptonization of relatively cold soft photons of the star off electrons of a hot shell of plasma of temperature T{sub e} of the order of 10{sup 9} K (or kT{sub e} {approx} 100 keV) that moves subrelativistically with the bulk velocity V{sub b} substantially less than the speed of light c. In this phase, the Comptonization parameter Y is high and the interaction between a blackbody-like soft seed photon population and hot electrons leads to formation of a saturated Comptonization spectrum modified by the subrelativistic bulk outflow. The second process is an upscattering of the previously Comptonized spectrum by the plasma outflow once it becomes relativistic. This process gives rise to the high-energy power-law (PL) component above the peak in the EF(E) diagram where F(E) is the energy flux. The latter process can be described by a convolution of the Comptonized spectrum with a broken-PL Green function. Possible physical scenarios for this second upscattering process are discussed. In the framework of our model, we give an interpretation of the Amati relation between the intrinsic spectral peak photon energy and radiated energy or luminosity, and we propose a possible explanation of the GRB temporal variability.

79 ASTRONOMY AND ASTROPHYSICS↗

Femtosecond pulse parameter estimation from photoelectron momenta using machine learning

Abstract Deep learning models have provided huge interpretation power for image-like data. Specifically, convolutional neural networks (CNNs) have demonstrated incredible acuity for tasks such as feature extraction or parameter estimation. Here we test CNNs on strong-field ionization photoelectron spectra, training on theoretical data sets to ‘invert’ experimental data. Pulse characterization is used as a ‘testing ground’, specifically we retrieve the laser intensity, where ‘traditional’ measurements typically lead to 20% uncertainty. We report on crucial data augmentation techniques required to successfully train on theoretical data and return consistent results from experiments, including accounting for detector saturation. The same procedure can be repeated to apply CNNs in a range of scenarios for strong-field ionization. Using a predictive uncertainty estimation, reliable laser intensity uncertainties of a few percent can be extracted, which are consistently lower than those given by traditional techniques. Using interpretability methods can reveal parts of the distribution that are most sensitive to laser intensity, which can be directly associated with holographic interferences. The CNNs employed provide an accurate and convenient ways to extract parameters, and represent a novel interpretational tool for strong-field ionization spectra.

Szołdra, Tomasz (ORCID:0000000228970506)↗

Solar system constraints on planetary Coriolis-type effects induced by rotation of distant masses

We phenomenologically put local constraints on the rotation of distant masses by using the planets of the solar system. First, we analytically compute the orbital secular precessions induced on the motion of a test particle about a massive primary by a Coriolis-like force, treated as a small perturbation, in the case of a constant angular velocity vector Ψ directed along a generic direction in space. The semimajor axis a and the eccentricity e of the test particle do not secularly change, contrary to the inclination I, the longitude of the ascending node Ω, the longitude of the pericenter varpi and the mean anomaly M. Then, we compare our prediction for (dot varpi) with the corrections Δdot varpi to the usual perihelion precessions of the inner planets recently estimated by fitting long data sets with different versions of the EPM ephemerides. We obtain as preliminary upper bounds |Ψ{sub z}| ≤ 0.0006−0.013 arcsec cty{sup −1}, |Ψ{sub x}| ≤ 0.1−2.7 arcsec cty{sup −1}, |Ψ{sub y}| ≤ 0.3−2.3 arcsec cty{sup −1}. Interpreted in terms of models of space-time involving cosmic rotation, our results are able to yield constraints on cosmological parameters like the cosmological constant Λ and the Hubble parameter H{sub 0} not too far from their values determined with cosmological observations and, in some cases, several orders of magnitude better than the constraints usually obtained so far from space-time models not involving rotation. In the case of the rotation of the solar system throughout the Galaxy, occurring clockwise about the North Galactic Pole, our results for Ψ{sub z} are in disagreement with the expected value of it at more than 3−σ level. Modeling the Oort cloud as an Einstein-Thirring slowly rotating massive shell inducing Coriolis-type forces inside yields unphysical results for its putative rotation.

79 ASTRONOMY AND ASTROPHYSICS↗

A POSSIBLE CARBON-RICH INTERIOR IN SUPER-EARTH 55 Cancri e

Terrestrial planets in the solar system, such as the Earth, are oxygen-rich, with silicates and iron being the most common minerals in their interiors. However, the true chemical diversity of rocky planets orbiting other stars is yet unknown. Mass and radius measurements are used to constrain the interior compositions of super-Earths (exoplanets with masses of 1-10 M{sub Circled-Plus }), and are typically interpreted with planetary interior models that assume Earth-centric oxygen-rich compositions. Using such models, the super-Earth 55 Cancri e (mass 8 M{sub Circled-Plus }, radius 2 R{sub Circled-Plus }) has been suggested to bear an interior composition consisting of Fe, silicates, and an envelope ({approx}> 10% by mass) of supercritical water. We report that the mass and radius of 55 Cancri e can also be explained by a carbon-rich solid interior made of Fe, C, SiC, and/or silicates and without a volatile envelope. While the data allow Fe mass fractions of up to 40%, a wide range of C, SiC, and/or silicate mass fractions are possible. A carbon-rich 55 Cancri e is also plausible if its protoplanetary disk bore the same composition as its host star, which has been reported to be carbon-rich. However, more precise estimates of the stellar elemental abundances and observations of the planetary atmosphere are required to further constrain its interior composition. The possibility of a C-rich interior in 55 Cancri e opens a new regime of geochemistry and geophysics in extraterrestrial rocky planets, compared to terrestrial planets in the solar system.

79 ASTRONOMY AND ASTROPHYSICS↗

Conformalized-KANs: Uncertainty Quantification with Coverage Guarantees for Kolmogorov-Arnold Networks (KANs) in Scientific Machine Learning

This paper explores uncertainty quantification (UQ) methods in the context of Kolmogorov–Arnold Networks (KANs). We apply an ensemble approach to KANs to obtain a heuristic measure of UQ, enhancing interpretability and robustness in modeling complex functions. Building on this, we introduce Conformalized-KANs, which integrate conformal prediction, a distribution-free UQ technique, with KAN ensembles to generate calibrated prediction intervals with guaranteed coverage.} Extensive numerical experiments are conducted to evaluate the effectiveness of these methods, focusing particularly on the robustness and accuracy of the prediction intervals under various hyperparameter settings. We show that the conformal KAN predictions can be applied to recent extensions of KANs, including Finite Basis KANs (FBKANs) and multifideilty KANs (MFKANs). The results demonstrate the potential of our approaches to significantly improve the reliability and applicability of KANs in scientific machine learning.

• Artificial intelligence (AI) / machine learning ↗

Photometric classification of Hyper Suprime-Cam transients using machine learning

Abstract The advancement of technology has resulted in a rapid increase in supernova (SN) discoveries. The Subaru/Hyper Suprime-Cam (HSC) transient survey, conducted from fall 2016 through spring 2017, yielded 1824 SN candidates. This gave rise to the need for fast type classification for spectroscopic follow-up and prompted us to develop a machine learning algorithm using a deep neural network with highway layers. This algorithm is trained by actual observed cadence and filter combinations such that we can directly input the observed data array without any interpretation. We tested our model with a dataset from the LSST classification challenge (Deep Drilling Field). Our classifier scores an area under the curve (AUC) of 0.996 for binary classification (SN Ia or non-SN Ia) and 95.3% accuracy for three-class classification (SN Ia, SN Ibc, or SN II). Application of our binary classification to HSC transient data yields an AUC score of 0.925. With two weeks of HSC data since the first detection, this classifier achieves 78.1% accuracy for binary classification, and the accuracy increases to 84.2% with the full dataset. This paper discusses the potential use of machine learning for SN type classification purposes.

Takahashi, Ichiro↗

Partial-wave decomposition of the Keldysh ionization amplitude

We present an alternative way of calculating the Keldysh amplitude, i.e., the length-gauge form of the ionization amplitude in the strong-field approximation. The amplitude is evaluated exactly by expanding it in Fourier components and partial waves. Comparisons of the semianalytic model predictions with results of ab initio numerical simulations of the time-dependent Schrödinger equation for the interaction of electrons in short-range potentials with intense laser light yield excellent agreement, for wavelengths from the single photon to the multiphoton to the tunneling regime. Specifically, for ionization from initial states with higher angular momentum quantum number, e.g., p states, a significant improvement over predictions based on the popular saddle-point approximation is found. Furthermore, the current model rate allows for interpretation of the strong-field ionization process in terms of multiphoton absorption pathways and angular momentum selection rules.

74 ATOMIC AND MOLECULAR PHYSICS↗

Linear- and circular-polarization photoionization angular distributions in H{sub 2} and H{sub 2}{sup +} by attosecond xuv laser pulses

Numerical solutions of the time-dependent Schroedinger equation, using a new unitary numerical method, are used to obtain gauge-independent photoionization angular distributions for two-dimensional (planar) models of aligned H{sub 2} and H{sub 2}{sup +} molecules in different electronic states. Simulations are performed with few-cycle attosecond (as) extreme-ultraviolet (xuv) laser pulses at wavelengths {lambda}= 10 nm [{omega}= 4.56 a.u. (atomic units), {tau}=2{pi}/{omega}= 1.38 a.u. = 33.4 as], and {lambda}= 5 nm ({omega}= 9.11 a.u., {tau}= 16.7 as) for linear and circular polarization ionizations at R{sub e}=1.675 a.u., where molecular orbital configurations dominate for which the electron wavelengths {lambda}{sub e{>=}}R{sub e} and at large internuclear distance R=10 a.u. ({lambda}{sub e}<R) where Heitler-London atomic configurations including ionic states H{sup +}H{sup -} are more appropriate. The simulations allow one to investigate the effects of electron correlation and entanglement in H{sub 2} at different R due to electron spin exchange as compared to the delocalized one-electron H{sub 2}{sup +} system. An ultrashort (delta function) pulse model is used to interpret interferences in momentum distributions.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Entanglement transitions from holographic random tensor networks

Here, we introduce a class of phase transitions separating quantum states with different entanglement features. An example of such an “entanglement phase transition” is provided by the many-body localization transition in disordered quantum systems, as it separates highly entangled thermal states at weak disorder from many-body localized states with low entanglement at strong disorder. In the spirit of random matrix theory, we describe a simple model for such transitions where a physical quantum many-body system lives at the “holographic” boundary of a bulk random tensor network. Using a replica trick approach, we map the calculation of the entanglement properties of the boundary system onto the free energy cost of fluctuating domain walls in a classical statistical mechanics model. This allows us to interpret transitions between volume-law and area-law scaling of entanglement as ordering transitions in this statistical mechanics model. Our approach allows us to get an analytic handle on the field theory of these entanglement transitions.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Thermoelectric power as a probe of density of states in correlated actinide materials: The case of PuCoGa 5 superconductor

We present measurements of the thermoelectric power of the plutonium-based unconventional superconductor PuCoGa 5 . The data is interpreted within a phenomenological model for the quasiparticle density of states of intermediate valence systems and the results are compared with results obtained from photoemission spectroscopy. The results are consistent with intermediate valence nature of 5f-electrons, furthermore, we propose that measurements of the Seebeck coefficient can be used as a probe of density of states in this material, thereby providing a link between transport measurements and photoemission in strongly correlated materials. Here, we discuss these results and their implications for the electronic structure determination of other strongly correlated systems, especially nuclear materials.

36 MATERIALS SCIENCE↗

Coriolis coupling effects in proton-pickup spectroscopic factors from 12 B

Spectroscopic factors to low-lying negative-parity states in Be 11 extracted from the B 12 ( d , He 3 ) Be 11 proton-removal reaction are interpreted within the rotational model. Earlier predictions of the p -wave proton-removal strengths in the strong-coupling limit of the Nilsson model underestimated the spectroscopic factors to the 3 / 2 1 - and 5 / 2 1 - states and suggested that deviations in the 1 + ground state of the odd-odd B 12 due to Coriolis coupling should be further explored. Furthermore, in this work we use the particle rotor model to take into account these effects and obtain a good description of the level scheme in B 11 , with a moderate K mixing of the proton Nilsson levels [110]1/2 and [101]3/2. This mixing, present in the 1 + bandhead of B 12 , is key to explaining the proton-pickup data.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

New collective structures in Au 179 and their implications for the triaxial deformation of the Pt 178 core

The extremely neutron-deficient isotope 179 Au has been studied by a combination of in-beam γ-ray and isomeric-decay spectroscopy. For in-beam spectroscopy, the recoil-isomer tagging technique was employed, using the known 3/2 – , T 1/2 = 328 ns isomer. A new rotational band, associated with the unfavored signature band of the 1⁢ℎ 9/2 ⊕2f⁢ 7/2 proton-intruder configuration, was revealed. A previously unknown, high-spin isomeric state with an excitation energy of 1743(17) keV and T 1/2 = 2.16⁢(8) ⁢µs was discovered. Five decay paths were identified, some of them feeding previously unknown non-yrast excited states, associated with the 1⁢i 13/2 proton-intruder configuration. Calculations based on the particle-plus-triaxial-rotor model were performed to interpret the data. On the basis of these calculations, the new 1⁢ℎ 9/2 ⊕2f 7/2 rotational band is interpreted as due to triaxial deformation of the underlying configuration with β 2 ≈ 0.26 and γ ≈ 27°. In conclusion, observed non-yrast states of the positive-parity 1i 13/2 intruder configuration are interpreted as due to triaxial deformation with β 2 ≈ 0.26 and γ ≈ 20°.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Reinvestigation of the level structures of the even-even nuclei 90 Zr and 92 Zr

Excited states of 90 Zr and 92 Zr were investigated using the fusion reactions 6 Li + 89 Y and 14 N + 82 Se , respectively. Based on the experimental data, 5 and 18 new 𝛾 rays have been added to the level schemes of 90 Zr and 92 Zr , respectively. The level structures of 90 Zr and 92 Zr have been interpreted with shell-model calculations using the GWBXG and SNET effective interactions. The neutron core-breaking and the importance of the 1⁢𝑔 7/2 and 1⁢ℎ 11/2 orbits for high-spin states of 92 Zr were discussed. Finally, the excitation energy of the neutron core-breaking of nuclei in the 𝐴=90 region was compared.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

g factors of coexisting isomeric states in {sup 188}Pb

The g factors of the 12{sup +}, 11{sup -}, and 8{sup -} isomeric states in {sup 188}Pb were measured using the time-differential perturbed angular distribution method as g(12{sup +})=-0.179(6), g(11{sup -})=+1.03(3), and g(8{sup -})=-0.037(7). The g factor of the 12{sup +} state follows the observed slight down-sloping evolution of the g factors of the i{sub 13/2}{sup 2} neutron spherical states with decreasing N. The g factors of the 11{sup -} and 8{sup -} isomers proposed as oblate and prolate deformed states, respectively, were interpreted within the rotational model, using calculated and empirical g factor values for the involved single-particle orbitals.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

High-spin level structure of {sup 115}Rh: Evolution of triaxiality in odd-even Rh isotopes

High-spin excited states in the neutron-rich nucleus {sup 115}Rh have been identified for the first time by studying prompt {gamma} rays from the spontaneous fission of {sup 252}Cf with the Gammasphere detector array. A new yrast band and a sideband are built in {sup 115}Rh. This level scheme is proposed to be built on the 7/2{sup +} ground state. The existence of a large signature splitting and an yrare band in {sup 115}Rh shows typical features of a triaxially deformed nucleus. The rigid triaxial rotor plus particle model is used to interpret the level structure of {sup 115}Rh. The level energies, the {gamma} branching ratios, the large signature splitting in the yrast band, and the inverted signature splitting in the yrare band in {sup 115}Rh are reproduced very well. Strong K mixing occurs in {sup 115}Rh at high spin.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Attenuation of {omega} mesons in cold nuclear matter

The attenuation of {omega} mesons in cold nuclear matter has been investigated via the time-dependent multiple-scattering Monte Carlo multicollisional (MCMC) intranuclear cascade model. The inelastic {omega} width deduced from CBELSA/TAPS Collaboration data of meson transparency in complex nuclei ({Gamma}*{approx_equal}30 MeV/c{sup 2}) is approximately 5 times lower than the value obtained with recent theoretical models and consistent with an in-medium total {omega}N cross section within 25-30 mb for an average meson momentum of 1.1 GeV/c . The momentum-dependent transparency ratios suggest an elastic/total cross-section ratio around 40%. For the case of CLAS Collaboration data a much higher width is deduced ({Gamma}* > or approx. 120 MeV/c{sup 2}), with the MCMC model providing a consistent interpretation of the data, assuming a much higher meson absorption ({sigma}{sub {omega}N}* > or approx. 100 mb) for p{sub {omega}{approx}1}.7 GeV/c.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗