Search NASA⌕ Search

SEARCH · Search NASA

Results for “generalization bounds”

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 253 records · Page 14
All sourcesNASA NTRSDOE OSTI
NASA NTRS2000

Predictive and Neural Predictive Control of Uncertain Systems

Accomplishments and future work are:(1) Stability analysis: the work completed includes characterization of stability of receding horizon-based MPC in the setting of LQ paradigm. The current work-in-progress includes analyzing local as well as global stability of the closed-loop system under various nonlinearities; for example, actuator nonlinearities; sensor nonlinearities, and other plant nonlinearities. Actuator nonlinearities include three major types of nonlineaxities: saturation, dead-zone, and (0, 00) sector. (2) Robustness analysis: It is shown that receding horizon parameters such as input and output horizon lengths have direct effect on the robustness of the system. (3) Code development: A matlab code has been developed which can simulate various MPC formulations. The current effort is to generalize the code to include ability to handle all plant types and all MPC types. (4) Improved predictor: It is shown that MPC design using better predictors that can minimize prediction errors. It is shown analytically and numerically that Smith predictor can provide closed-loop stability under GPC operation for plants with dead times where standard optimal predictor fails. (5) Neural network predictors: When neural network is used as predictor it can be shown that neural network predicts the plant output within some finite error bound under certain conditions. Our preliminary study shows that with proper choice of update laws and network architectures such bound can be obtained. However, much work needs to be done to obtain a similar result in general case.

Kelkar, Atul G.↗
NASA NTRS2000

Any Two Learning Algorithms Are (Almost) Exactly Identical

This paper shows that if one is provided with a loss function, it can be used in a natural way to specify a distance measure quantifying the similarity of any two supervised learning algorithms, even non-parametric algorithms. Intuitively, this measure gives the fraction of targets and training sets for which the expected performance of the two algorithms differs significantly. Bounds on the value of this distance are calculated for the case of binary outputs and 0-1 loss, indicating that any two learning algorithms are almost exactly identical for such scenarios. As an example, for any two algorithms A and B, even for small input spaces and training sets, for less than 2e(-50) of all targets will the difference between A's and B's generalization performance of exceed 1%. In particular, this is true if B is bagging applied to A, or boosting applied to A. These bounds can be viewed alternatively as telling us, for example, that the simple English phrase 'I expect that algorithm A will generalize from the training set with an accuracy of at least 75% on the rest of the target' conveys 20,000 bytes of information concerning the target. The paper ends by discussing some of the subtleties of extending the distance measure to give a full (non-parametric) differential geometry of the manifold of learning algorithms.

Wolpert, David H.↗
NASA NTRS1987

Robust time and frequency domain estimation methods in adaptive control

A robust identification method was developed for use in an adaptive control system. The type of estimator is called the robust estimator, since it is robust to the effects of both unmodeled dynamics and an unmeasurable disturbance. The development of the robust estimator was motivated by a need to provide guarantees in the identification part of an adaptive controller. To enable the design of a robust control system, a nominal model as well as a frequency-domain bounding function on the modeling uncertainty associated with this nominal model must be provided. Two estimation methods are presented for finding parameter estimates, and, hence, a nominal model. One of these methods is based on the well developed field of time-domain parameter estimation. In a second method of finding parameter estimates, a type of weighted least-squares fitting to a frequency-domain estimated model is used. The frequency-domain estimator is shown to perform better, in general, than the time-domain parameter estimator. In addition, a methodology for finding a frequency-domain bounding function on the disturbance is used to compute a frequency-domain bounding function on the additive modeling error due to the effects of the disturbance and the use of finite-length data. The performance of the robust estimator in both open-loop and closed-loop situations is examined through the use of simulations.

Lamaire, Richard Orville↗
DOE OSTI2025

Comparing approaches for introducing polycyclic aromatic hydrocarbons to Ge(001) to seed graphene nanoribbon synthesis by CH 4 chemical vapor deposition

Here, we evaluate two approaches for introducing polycyclic aromatic hydrocarbons (PAHs) to a graphene catalyst substrate as part of a two-step chemical vapor deposition (CVD) process for growing graphene nanoribbons (GNRs). In this process, PAHs first form graphene-like seeds on Ge, and then GNRs are subsequently evolved from these PAH-derived seeds via substrate-mediated anisotropic growth kinetics during the CVD of CH4. The first seeding approach sublimes controlled doses of PAH thin films into the CVD chamber, while the second delivers PAHs directly from the vapor phase at set concentrations. Using these two approaches, we measure the dependence of GNR density on PAH dose and compare the experimental results with predictions from a rate model of PAH diffusion, desorption, and clustering. We find that PAHs that more strongly adsorb to the catalyst surface (generally larger PAHs) desirably remain more individualized prior to GNR evolution whereas smaller, more weakly bound PAHs aggregate into larger clusters with various sizes on Ge that are undesirable for synthesizing more monodisperse GNRs. As a result, this work is important because it offers a framework that enables the rational selection and design of seed molecules, advancing anisotropic GNR CVD synthesis.

CVD↗
DOE OSTI2024

Production of meson molecules in ultraperipheral heavy ion collisons

In this work we present a calculation of exotic charmonium production in ultraperipheral collisions, in which the exotic state is explicitly treated as a meson molecule. Our formalism is general, but we focus on the lightest possible exotic charmonium state: a 𝐷 + ⁢𝐷 − molecular bound state. It was proposed some time ago, and it has been an object of experimental searches. Here we study the production of the open charm pair in the process 𝛾⁢𝛾 → 𝐷 + ⁢𝐷 − . Then we use a prescription to project the free pair |𝐷 + ⁢𝐷 − ⟩ onto a bound state at the amplitude level and compute the cross section of the process 𝛾⁢𝛾 → 𝐵 (where 𝐵 is the bound state). Finally, we convolute this last cross section with the equivalent photon distributions coming from the projectile and target in an ultraperipheral collision and find the 𝐴⁢𝐴 → 𝐴⁢𝐴⁢𝐵 cross section, which, for 𝑃⁢𝑏 −𝑃⁢𝑏 collisions at $\sqrt{s_{NN}}$ =5.02 TeV, is of the order of 3 μ⁢b.

Sobrinho, F. C.↗
DOE OSTI2025

Warm and fuzzy dark matter: Free streaming of wave dark matter

Wave or fuzzy dark matter that is produced with relativistic wave numbers exhibits free-streaming effects analogous to warm or hot particle dark matter with relativistic momenta. Axions produced after inflation provide such a warm or mildly relativistic candidate, where the enhanced suppression and observational bounds are only moderately stronger than that from wave propagation of initially cold axions. More generally, the free-streaming damping also impacts isocurvature fluctuations from generation in causally disconnected patches. As coherent spatial fluctuations free stream away they leave incoherent and transient superpositions in their wakes. These multiple wave momentum streams are the wave analog of particle phase space fluctuations or directional collisionless damping of massive neutrinos or hot dark matter. The observable impact on both adiabatic and isocurvature fluctuations of fuzzy dark matter can differ from their cold dark matter counterparts due to free streaming depending on how warm or hot is their momentum distribution.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2025

Non-Gaussian Generalized Two-Mode Squeezing: Applications to Two-Ensemble Spin Squeezing and Beyond

Bosonic two-mode squeezed states are paradigmatic entangled Gaussian states that have wide utility in quantum information and metrology. Here, in this study, we show that the basic structure of these states can be generalized to arbitrary bipartite quantum systems in a manner that allows simultaneous, Heisenberg-limited estimation of two independent parameters for finite-dimensional systems. Further, we show that these general states can always be stabilized by a relatively simple Markovian dissipative process. In the specific case where the two subsystems are ensembles of two-level atoms or spins, our generalized states define a notion of two-mode spin squeezing that is valid beyond the Gaussian limit and that enables true multiparameter estimation. We discuss how generalized Ramsey measurements allow one to reach the two-parameter quantum Cramér-Rao bound, and how the dissipative preparation scheme is compatible with current experiments.

Mamaev, Mikhail [Univ. of Chicago, IL (United Stat↗
NASA NTRS1979

A generalized modal shock spectra method for spacecraft loads analysis

Unlike an earlier shock spectra approach, generalization permits an accurate elastic interaction between the spacecraft and launch vehicle to obtain accurate bounds on the spacecraft response and structural loads. In addition, the modal response from a previous launch vehicle transient analysis with or without a dummy spacecraft - is exploited to define a modal impulse as a simple idealization of the actual forcing function. The idealized modal forcing function is then used to derive explicit expressions for an estimate of the bound on the spacecraft structural response and forces. Greater accuracy is achieved with the present method over the earlier shock spectra, while saving much computational effort over the transient analysis.

Trubert, M.↗
NASA NTRS1979

Semiclassical theory of electronically nonadiabatic transitions in molecular collision processes

An introductory account of the semiclassical theory of the S-matrix for molecular collision processes is presented, with special emphasis on electronically nonadiabatic transitions. This theory is based on the incorporation of classical mechanics with quantum superposition, and in practice makes use of the analytic continuation of classical mechanics into the complex space of time domain. The relevant concepts of molecular scattering theory and related dynamical models are described and the formalism is developed and illustrated with simple examples - collinear collision of the A+BC type. The theory is then extended to include the effects of laser-induced nonadiabatic transitions. Two bound continuum processes collisional ionization and collision-induced emission also amenable to the same general semiclassical treatment are discussed.

Lam, K. S.↗
NASA NTRS1993

Tidal disruptions by supermassive black holes - Hydrodynamic evolution of stars on a Schwarzschild background

We present a three-dimensional numerical study of tidal disruption of a main-sequence star by a supermassive black hole. The simulations include general relativistic effects which are important in this regime. We analyze stars in a marginally bound orbit around the black hole with pericentric separation of a few Schwarzschild radii. We show that during a close passage, as a result of relativistic effects analogous to the perihelion shift, the trajectories of the debris of the star fan out into a crescent-like shape centered on the black hole. We also discuss the increase of the central density of the star as it approaches pericentric distance, the fraction of the debris accreted by the hole, its accretion rate, the distribution of debris orbits bound to the hole, and the velocity of unbound ejected material. We compare these results with the disruption of the star by a Newtonian point mass.

Laguna, Pablo↗
NASA NTRS2014

Far-Reaching Impacts of African Dust- A Calipso Perspective

African dust can transport across the tropical Atlantic and reach the Amazon basin, exerting far-reaching impacts on climate in downwind regions. The transported dust influences the surface-atmosphere interactions and cloud and precipitation processes through perturbing the surface radiative budget and atmospheric radiative heating and acting as cloud condensation nuclei and ice nuclei. Dust also influences biogeochemical cycle and climate through providing nutrients vital to the productivity of ocean biomass and Amazon forests. Assessing these climate impacts relies on an accurate quantification of dust transport and deposition. Currently model simulations show extremely large diversity, which calls for a need of observational constraints. Kaufman et al. (2005) estimated from MODIS aerosol measurements that about 144 Tg of dust is deposited into the tropical Atlantic and 50 Tg of dust into the Amazon in 2001. This estimated dust import to Amazon is a factor of 3-4 higher than other observations and models. However, several studies have argued that the oversimplified characterization of dust vertical profile in the study would have introduced large uncertainty and very likely a high bias. In this study we quantify the trans-Atlantic dust transport and deposition by using 7 years (2007-2013) observations from CALIPSO lidar. CALIPSO acquires high-resolution aerosol extinction and depolarization profiles in both cloud-free and above-cloud conditions. The unique CALIPSO capability of profiling aerosols above clouds offers an unprecedented opportunity of examining uncertainties associated with the use of MODIS clear-sky data. Dust is separated from other types of aerosols using the depolarization measurements. We estimated that on the basis of 7-year average, 118142 Tg of dust is deposited into the tropical Atlantic and 3860 Tg of dust into the Amazon basin. Substantial interannual variations are observed during the period, with the maximum to minimum ratio of about 1.6 and 2.5 for the deposition to the tropical Atlantic and Amazon, respectively. The MODIS-based estimates appear to fall within the range of CALIPSO-based estimates; and the difference between MODIS and CALIPSO estimates can be largely attributed to the interannual variability, which is corroborated by long-term surface dust concentration observations in the tropical Atlantic. Considering that CALIPSO generally tends to underestimate the aerosol loading, our estimate is likely to represent a low bound for the dust transport and deposition estimate. The finding suggests that models have substantial biases and considerable effort is needed to improve model simulations of dust cycle.

dust↗
NASA NTRS1997

A Near-Wall Reynolds-Stress Closure Without Wall Normals

Turbulent wall-bounded complex flows are commonly encountered in engineering practice and are of considerable interest in a variety of industrial applications. The presence of a wall significantly affects turbulence characteristics. In addition to the wall effects, turbulent wall-bounded flows become more complicated by the presence of additional body forces (e.g. centrifugal force and Coriolis force) and complex geometry. Most near-wall Reynolds stress models are developed from a high-Reynolds-number model which assumes turbulence is homogenous (or quasi-homogenous). Near-wall modifications are proposed to include wall effects in near-wall regions. In this process, wall normals are introduced. Good predictions could be obtained by Reynolds stress models with wall normals. However, ambiguity arises when the models are applied in flows with multiple walls. Many models have been proposed to model turbulent flows. Among them, Reynolds stress models, in which turbulent stresses are obtained by solving the Reynolds stress transport equations, have been proved to be the most successful ones. To apply the Reynolds stress models to wall-bounded flows, near-wall corrections accounting for the wall effects are needed, and the resulting models are called near-wall Reynolds stress models. In most of the existing near-wall models, the near-wall corrections invoke wall normals. These wall-dependent near-wall models are difficult to implement for turbulent flows with complex geometry and may give inaccurate predictions due to the ambiguity of wall normals at corners connecting multiple walls. The objective of this study is to develop a more general and flexible near-wall Reynolds stress model without using any wall-dependent variable for wall-bounded turbulent flows. With the aid of near-wall asymptotic analysis and results of direct numerical simulation, a new near-wall Reynolds stress model (NNWRS) is formulated based on Speziale et al.'s high-Reynolds-stress model with wall-independent near-wall corrections. Moreover, only one damping function is used for flows with a wide range of Reynolds numbers to ensure that the near-wall modifications diminish away from the walls.

Yuan, S. P.↗
NASA NTRS1986

Chaos and nonlinear dynamics of single-particle orbits in a magnetotaillike magnetic field

The properties of charged-particle motion in Hamiltonian dynamics are studied in a magnetotaillike magnetic field configuration. It is shown by numerical integration of the equation of motion that the system is generally nonintegrable and that the particle motion can be classified into three distinct types of orbits: bounded integrable orbits, unbounded stochastic orbits, and unbounded transient orbits. It is also shown that different regions of the phase space exhibit qualitatively different responses to external influences. The concept of 'differential memory' in single-particle distributions is proposed. Physical implications for the dynamical properties of the magnetotail plasmas and the possible generation of non-Maxwellian features in the distribution functions are discussed.

Chen, J.↗
NASA NTRS1992

Structure of merger remnants. I - Bulgeless progenitors

The study examines mergers of identical galaxies consisting of self-gravitating disks and halos in the context of the suggestion that such events may form elliptical galaxies. It is shown that the luminous remnants of such mergers do indeed share many common properties with observed ellipticals. Specifically, the end states of the simulations considered rotate slowly in regions of relatively high surface density, having typical values of less than about 0.2 there. Morphologically, the remnants display a variety of structures, including shells and loops comprising loosely bound material and boxy and disky isophotes. The luminous matter is well-fitted by ellipsoidal generalizations of Hernquists's (1990, 1992) model for elliptical galaxies, implying that the surface brightness profiles are essentially de Vaucouleurs-like over a large radial interval. It is proposed that mergers of pure stellar disks do not represent an attractive mechanism for the production of massive elliptical galaxies.

Hernquist, Lars↗
NASA NTRSDate not supplied

Seeing is Believing: Monitoring Future Time Temporal Logic

Runtime monitors for future-time unbounded temporal logics like RVLTL, LTL 3 and FLTL, have double-exponential (2^2^n) worst-case space complexity bounds in size of the input formula. The semantics of these logics require monitors to perform general satisfiability solving for LTL expressions, a well-studied problem whose computational complexity is NP-hard and PSPACE-complete. This paper introduces an unbounded future-time linear temporal logic defined over a lattice. We call our logic an incremental temporal logic as it can be viewed as incrementally constructing proofs about the trace. On this account, we view online runtime monitoring as a decision procedure for proofs systems about incrementally growing traces. We demonstrate that our incremental temporal logic allows monitor construction to void satisfiability solving while still soundly detecting when the property is violated in an online fashion. This enables asymptotic improvements in space complexity. As proof, we provide a procedure to construct monitors that utilize linear space and time in the size of the input formula, while remaining constant in the size of the input stream and suitable for online monitoring. We further demonstrate, through several examples, that our incremental temporal logic is straightforward to adopt and practical for runtime verification.

temporal logic↗
DOE OSTI2024

Compressing branch-and-bound trees

A branch-and-bound (BB) tree certifies a dual bound on the value of an integer program. In this work, we introduce the tree compression problem (TCP): Given a BB tree T that certifies a dual bound, can we obtain a smaller tree with the same (or stronger) bound by either (1) applying a different disjunction at some node in T or (2) removing leaves from T? Here we believe such post-hoc analysis of BB trees may assist in identifying helpful general disjunctions in BB algorithms. We initiate our study by considering computational complexity and limitations of TCP. We then conduct experiments to evaluate the compressibility of realistic branch-and-bound trees generated by commonly-used branching strategies, using both an exact and a heuristic compression algorithm.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2008

Sparse Regression as a Sparse Eigenvalue Problem

We extend the l0-norm "subspectral" algorithms for sparse-LDA [5] and sparse-PCA [6] to general quadratic costs such as MSE in linear (kernel) regression. The resulting "Sparse Least Squares" (SLS) problem is also NP-hard, by way of its equivalence to a rank-1 sparse eigenvalue problem (e.g., binary sparse-LDA [7]). Specifically, for a general quadratic cost we use a highly-efficient technique for direct eigenvalue computation using partitioned matrix inverses which leads to dramatic x103 speed-ups over standard eigenvalue decomposition. This increased efficiency mitigates the O(n4) scaling behaviour that up to now has limited the previous algorithms' utility for high-dimensional learning problems. Moreover, the new computation prioritizes the role of the less-myopic backward elimination stage which becomes more efficient than forward selection. Similarly, branch-and-bound search for Exact Sparse Least Squares (ESLS) also benefits from partitioned matrix inverse techniques. Our Greedy Sparse Least Squares (GSLS) generalizes Natarajan's algorithm [9] also known as Order-Recursive Matching Pursuit (ORMP). Specifically, the forward half of GSLS is exactly equivalent to ORMP but more efficient. By including the backward pass, which only doubles the computation, we can achieve lower MSE than ORMP. Experimental comparisons to the state-of-the-art LARS algorithm [3] show forward-GSLS is faster, more accurate and more flexible in terms of choice of regularization

Exact Sparse Least Squares (ESLS)↗
DOE OSTI2025

McCormick envelopes in mixed-integer PDE-constrained optimization

McCormick envelopes are a standard tool for deriving convex relaxations of optimization problems that involve polynomial terms. Such McCormick relaxations provide lower bounds, for example, in branch-and-bound procedures for mixed-integer nonlinear programs but have not gained much attention in PDE-constrained optimization so far. This lack of attention may be due to the distributed nature of such problems, which on the one hand leads to infinitely many linear constraints (generally state constraints that may be difficult to handle) in addition to the state equation for a pointwise formulation of the McCormick envelopes and renders bound-tightening procedures that successively improve the resulting convex relaxations computationally intractable. We analyze McCormick envelopes for a model problem class that is governed by a semilinear PDE involving a bilinearity and integrality constraints. We approximate the nonlinearity and in turn the McCormick envelopes by averaging the involved terms over the cells of a partition of the computational domain on which the PDE is defined. This yields convex relaxations that underestimate the original problem up to an a priori error estimate that depends on the mesh size of the discretization. These approximate McCormick relaxations can be improved by means of an optimization-based bound-tightening procedure. We show that their minimizers converge to minimizers to a limit problem with a pointwise formulation of the McCormick envelopes when driving the mesh size to zero. We provide a computational example, for which we certify all of our imposed assumptions. The results point to both the potential of the methodology and the gaps in the research that need to be closed. Our methodology provides a framework first for obtaining pointwise underestimators for nonconvexities and second for approximating them with finitely many linear inequalities in an infinite-dimensional setting.

Approximations and Expansions↗