Search

Autonomous selection of optimal options for data collection from multiple alternatives is challenging in uncertain environments. When secondary information about options is accessible, such problems can be framed as contextual multi-armed bandits (CMABs). Neuro-inspired active inference has gained interest for its ability to balance exploration and exploitation using the expected free energy objective function. Unlike previous studies that showed the effectiveness of active inference based strategy for CMABs using synthetic data, this study aims to apply active inference to realistic scenarios, using a simulated mineralogical survey site selection problem. Hyperspectral data from AVIRIS-NG at Cuprite, Nevada, serves as contextual information for predicting outcome probabilities, while geologists' mineral labels represent outcomes. Monte Carlo simulations assess the robustness of active inference against changing expert preferences. Results show that active inference requires fewer iterations than standard bandit approaches with real-world noisy and biased data, and performs better when outcome preferences vary online by adapting the selection strategy to align with expert shifts., Comment: 11 pages, 12 figures, submitted to IEEE Transactions on Robotics

In this work, we investigate the electrical conductivity of carbon nanotubes (CNTs), with a particular focus on the effects of doping. Using a first-principles approach, we study the electronic structure, phonon dispersion, and electron-phonon scattering to understand the finite-temperature electrical transport properties in CNTs. Our study covers both prototypical metallic and semiconducting CNTs, with special emphasis on the influence of typical defects such as vacancies and the incorporation of copper or nitrogen, such as pyridinic N, pyrrolic N, graphitic N, and oxidized N. Our theoretical study shows significant improvements in the electrical conduction properties of copper-CNT composites, especially when semiconducting CNTs are functionalized with nitrogen. Doping is found to cause significant changes in the electronic density of states near the Fermi level, which affects the electrical conductivity. Calculations show that certain types of functional groups, such as N-pyrrolic, result in a ~30-fold increase in the conductivity of semiconducting CNTs compared to Cu-incorporated CNTs alone. For metallic CNTs, the conductivity is in agreement with existing experimental data, and our prediction of significant increases in conductivity with N-pyrrolic functional group is consistent with recent experimental results, demonstrating the effectiveness of doping in modifying conductivity. Our study provides valuable insight into the electronic properties of doped CNTs and contributes to the development of ultra-high conductivity CNT composites., Comment: 16 pages, 4 figures, 39 references

Europa is an icy ocean world, differentiated into a floating ice shell and solid interior, separated by a global ocean. The classical spin-orbit coupling problem considers a satellite as a single rigid body, but in the case of Europa, the existence of the subsurface ocean enables independent motion of the ice shell and solid interior. This paper explores the spin-orbit coupling problem for Europa from a dynamical perspective, yielding illuminating analytical and numerical results. We determine that the spin behavior of Europa is influenced by processes not captured by the classical single rigid body spin-orbit coupling analysis. The tidal locking process for Europa is governed by the strength of gravity-gradient coupling between the ice shell and solid interior, with qualitatively different behavior depending on the scale of this effect. In this coupled rigid model, the shell can potentially undergo large angular displacements from the solid interior, and the coupling plays an outsize role in the dynamical evolution of the moon, even without incorporating the dissipative effects of shell non-rigidity. We additionally discuss the effects of a realistic viscoelastic shell, and catalogue other torques that we expect to be sub-dominant in Europa's spin dynamics, or whose importance is unknown. Finally, we explore how the choice of tidal model affects the resulting equilibrium spin state., Comment: Revision 2 - fixing typos, editing figures, and adding references and acknowledgements. 38 pages total with 21 figures

Linear classifier probes are frequently utilized to better understand how neural networks function. Researchers have approached the problem of determining unit importance in neural networks by probing their learned, internal representations. Linear classifier probes identify highly selective units as the most important for network function. Whether or not a network actually relies on high selectivity units can be tested by removing them from the network using ablation. Surprisingly, when highly selective units are ablated they only produce small performance deficits, and even then only in some cases. In spite of the absence of ablation effects for selective neurons, linear decoding methods can be effectively used to interpret network function, leaving their effectiveness a mystery. To falsify the exclusive role of selectivity in network function and resolve this contradiction, we systematically ablate groups of units in subregions of activation space. Here, we find a weak relationship between neurons identified by probes and those identified by ablation. More specifically, we find that an interaction between selectivity and the average activity of the unit better predicts ablation performance deficits for groups of units in AlexNet, VGG16, MobileNetV2, and ResNet101. Linear decoders are likely somewhat effective because they overlap with those units that are causally important for network function. Interpretability methods could be improved by focusing on causally important units., Comment: 6 pages, 3 figures, to be published in I Can't Believe It's Note Better Workshop at NeurIPS 2022

We present an empirical dataset surveying the deep learning phenomenon on fully-connected feed-forward multilayer perceptron neural networks. The dataset, which is now freely available online, records the per-epoch training and generalization performance of 483 thousand distinct hyperparameter choices of architectures, tasks, depths, network sizes (number of parameters), learning rates, batch sizes, and regularization penalties. Repeating each experiment an average of 24 times resulted in 11 million total training runs and 40 billion epochs recorded. Accumulating this 1.7 TB dataset utilized 11 thousand CPU core-years, 72.3 GPU-years, and 163 node-years. In surveying the dataset, we observe durable patterns persisting across tasks and topologies. We aim to spark scientific study of machine learning techniques as a catalyst for the theoretical discoveries needed to progress the field beyond energy-intensive and heuristic practices.

We are preparing a telescope system to carry out a survey of detection and analysis of lunar impact flashes. In the framework of this project, here we present all necessary methods to automatically identify these luminous events, their lunar impact coordinates, the origin of the impacting meteoroids, as well as the estimation of their physical properties such as mass and size. We tested our methods against confirmed impact events and constructed the meteoroid size frequency distribution of impactors using literature data of the last 20 years. In addition, we present the first lunar impact event observed from the Observatoire de la Cote d'Azur that was detected during the testing phase of our project.

All LIDAR instruments are not the same, and advancement of LIDAR technology requires an ongoing interest and demand from the community to foster further development of the required components. The purpose of this paper is to make the community aware of the need for further technical development, and the potential payoff of investing experimental time, money and thought into the next generation of LIDARs., Comment: 12 pages, 1 figure. arXiv admin note: substantial text overlap with arXiv:1406.0030

Wind is the process that connects Mars' climate system. Measurements of Mars atmospheric winds from orbit would dramatically advance our understanding of Mars and help prepare for human exploration of the Red Planet. Multiple instrument candidates are in development and will be ready for flight in the next decade. We urge the Decadal Survey to make these measurements a priority for 2023-2032., Comment: A White Paper submitted to the Planetary Science and Astrobiology Decadal Survey 2023-2032

Water ice is thought to be trapped in large permanently shadowed regions (PSRs) in the Moon's polar regions, due to their extremely low temperatures. Here, we show that many unmapped cold traps exist on small spatial scales, substantially augmenting the areas where ice may accumulate. Using theoretical models and data from the Lunar Reconnaissance Orbiter, we estimate the contribution of shadows on scales from 1 km down to 1 cm, the smallest distance over which we find cold-trapping to be effective for water ice. Approximately 10-20\% of the permanent cold trap area for water is found to be contained in these "micro cold traps," which are the most numerous cold traps on the Moon. Consideration of all spatial scales therefore substantially increases the number of cold traps over previous estimates, for a total area of ~40,000 km^2. A majority of cold traps for water ice is found at latitudes >80{\deg} because permanent shadows equatorward of 80{\deg} are typically too warm to support ice accumulation. Our results show that water trapped at the lunar poles may be more accessible as a resource for future missions than previously thought.

Europa's surface reflectance exhibits a pronounced hemispheric dichotomy, which is hypothesized to form due to enhanced irradiation of the trailing hemisphere by energetic particles entrained in the jovian magnetosphere. We propose that this pattern can only persist if the timescale for discoloration is much shorter than that of Europa's rotation relative to the synchronous state, and provide a means for constraining the rotation rate using the observed color pattern. By decomposing the longitudinal ultraviolet and visible color variations from Voyager data into sine and cosine terms, we find no detectable signature of non-synchronous rotation (NSR). This same conclusion is reached with two observational models of discoloration: one representing an actively discoloring surface, and the other assuming that the present-day exogenic discoloration on the surface is in steady-state. Magnitudes of the expected signature are presented as functions of the age of the crater Pwyll, which is used to constrain the timescale of discoloration. Furthermore, we develop a physical model of discoloration to validate the geometric models, producing consistent results. The failure to identify a signature of NSR using Europa's hemispheric color dichotomy magnifies the outstanding problem of the origin of the stress to explain Europa's pervasive tectonic features.

We present a data partitioning technique performed over skip graphs that promotes significant quantitative and qualitative improvements on NUMA locality in concurrent data structures, as well as reduced contention. We build on previous techniques of thread-local indexing and laziness, and, at a high level, our design consists of a partitioned skip graph, well-integrated with thread-local sequential maps, operating without contention. As a proof-of-concept, we implemented map and relaxed priority queue ADTs using our technique. Maps were conceived using lazy and non-lazy approaches to insertions and removals, and our implementations are shown to be competitive with state-of-the-art maps. We observe a 6x higher CAS locality, a 68.6% reduction on the number of remote CAS operations, and a increase from 88.3% to 99% CAS success rate when using a lazy skip graph as compared to a control skip list (subject to the same codebase, optimizations, and implementation practices). Qualitatively speaking, remote memory accesses are not only reduced in number, but the larger the NUMA distance between threads, the larger the reduction is. We consider two alternative implementations of relaxed priority queues that further take advantage of our data partitioning over skip graphs: (a) using ``spraying'', a well-known random-walk technique usually performed over skip lists, but now performed over skip graphs; and (b) a custom protocol that traverses the skip graph deterministically, marking elements along this traversal. We provide formal arguments indicating that the first approach is more \emph{relaxed}, that is, that the span of removed keys is larger, while the second approach has smaller contention. Experimental results indicate that the approach based on spraying performs better on skip graphs, yet both seem to scale appropriately.

Excitonic complexes in type-II quantum-ring heterostructures may be considered as artificial atoms due to the confinement of only one charge-carrier type in an artificial nucleus. Binding energies of excitons, trions, and biexcitons in these nanostructures are then effectively ionization energies of these artificial atoms. The binding energies reported here are calculated within the effective-mass approximation using the diffusion quantum Monte Carlo method and realistic geometries for gallium antimonide rings in gallium arsenide. The electrons form a halo outside the ring, with very little charge density inside the central cavity of the ring. The de-excitonization and binding energies of the complexes are relatively independent of the precise shape of the ring.

Kuiper belt object (KBO) Eris is exceptionally bright with a greater visible geometric albedo than any other known KBO. Its infrared reflectance spectrum is dominated by methane, which should form tholins that darken the surface on timescales much shorter than the age of the Solar System. Thus one or more ongoing processes probably maintain its brightness. Eris is predicted to have a primarily nitrogen atmosphere that is in vapor pressure equilibrium with nitrogen-ice and is collisional (not ballistic). Eris's eccentric orbit is expected to result in two atmospheric regimes: (1) a period near perihelion when the atmosphere is global (analogous to the atmospheres of Mars, Triton, and Pluto) and (2) a period near aphelion when only a local atmosphere exists near the warmest region (analogous to the atmosphere of Io). A numerical model developed to simulate Eris's thermal and volatile evolution in the local atmosphere regime is presented. The model conserves energy, mass, and momentum while maintaining vapor pressure equilibrium. It is adaptable to other local, collisional, sublimation atmospheres, which in addition to Io and Eris, may occur on several volatile-bearing KBOs. The model was applied for a limiting case where Eris is fixed at aphelion and has an initial nitrogen-ice mass everywhere equal to the precipitable column of nitrogen in Pluto's atmosphere during the New Horizons encounter (the resultant mass if the Pluto atmosphere collapsed uniformly onto the surface). The model results indicate that (1) transport of nitrogen in the local, collisional, sublimation atmosphere regime is significant, (2) changes of Eris's albedo or color from nitrogen transport may be observable, and (3) uniform collapse of a global, nitrogen atmosphere likely cannot explain Eris's anomalous albedo in the present epoch. Seasonal volatile transport remains a plausible hypothesis to explain..., Comment: Accepted for publication in Icarus

We used infrared data from the Lunar Reconnaissance Orbiter (LRO) Diviner Lunar Radiometer Experiment to globally map thermophysical properties of the Moon's regolith fines layer. Thermal conductivity varies from 7.4$\times$10$^{-4}$ W m$^{-1}$ K$^{-1}$ at the surface, to 3.4$\times$10$^{-3}$ W m$^{-1}$ K$^{-1}$ at depths of $\sim$1 m, given density values of 1100 kg m$^{-3}$ at the surface, to 1800 kg m$^{-3}$ at 1-m depth. On average, the scale height of these profiles is $\sim$7 cm, corresponding to a thermal inertia of 55 $\pm$2 J m$^{-2}$ K$^{-1}$ s$^{-1/2}$ at 273 K, relevant to the diurnally active near-surface layer, $\sim$4-7 cm. The temperature-dependence of thermal conductivity and heat capacity leads to a $\sim$2$\times$ diurnal variation in thermal inertia at the equator. On global scales, the regolith fines are remarkably uniform, implying rapid homogenization by impact gardening of this layer on timescales $<$ 1 Gyr. Regional and local scale variations show prominent impact features $<$ 1 Gyr old, including higher thermal inertia ($>$ 100 J m$^{-2}$ K$^{-1}$ s$^{-1/2}$) in the interiors and ejecta of Copernican-aged impact craters, and lower thermal inertia ($<$ 50 J m$^{-2}$ K$^{-1}$ s$^{-1/2}$) within the lunar cold spots identified by Bandfield et al. (2014). Observed trends in ejecta thermal inertia provide a potential tool for age-dating craters of previously unknown age, complementary to the approach suggested by Ghent et al. (2014). Several anomalous regions are identified in the global 128 pixels-per-degree maps presented here, including a high-thermal inertia deposit near the antipode of Tycho crater., Comment: Accepted for publication in Journal of Geophysical Research - Planets

We present a concept for using a polarization sensitive multispectral lidar to map the seasonal distribution and exchange of volatiles among the reservoirs of the Martian surface and atmosphere. The LIDAR instrument will be a multi-wavelength, altitude-resolved, active near-infrared (NIR, with 10 bands around 1.6 microns) instrument to measure the reflected intensity and polarization of backscattered radiation from planetary surfaces and atmospheres. The proposed instrument would be ideally suited for a mission to Mars to comprehensively investigate the nature and seasonal distributions of volatiles and aerosols. The investigation would include the abundance of atmospheric dust and condensed volatiles, surface and cloud/aerosol grain sizes and shapes, ice and dust particle microphysics and also variations in atmospheric chemistry during multiple overflight local times throughout polar night and day. Such an instrument would be ideal for mapping and detection of recently detected CO2 frost phenomena and H2O and CO2 precipitation events in the polar regions of Mars. Herein we discuss the applicability of this instrument to detect and map sublimation/deposition 'mode flips' recently discovered by Brown et al. (2016) using the CRISM passive infrared sensor on Mars Reconnaissance Orbiter., Comment: 9 pages, 3 figures

To enable safe and efficient human-robot collaboration in shared workspaces it is important for the robot to predict how a human will move when performing a task. While predicting human motion for tasks not known a priori is very challenging, we argue that single-arm reaching motions for known tasks in collaborative settings (which are especially relevant for manufacturing) are indeed predictable. Two hypotheses underlie our approach for predicting such motions: First, that the trajectory the human performs is optimal with respect to an unknown cost function, and second, that human adaptation to their partner's motion can be captured well through iterative re-planning with the above cost function. The key to our approach is thus to learn a cost function which "explains" the motion of the human. To do this, we gather example trajectories from pairs of participants performing a collaborative assembly task using motion capture. We then use Inverse Optimal Control to learn a cost function from these trajectories. Finally, we predict reaching motions from the human's current configuration to a task-space goal region by iteratively re-planning a trajectory using the learned cost function. Our planning algorithm is based on the trajectory optimizer STOMP, it plans for a 23 DoF human kinematic model and accounts for the presence of a moving collaborator and obstacles in the environment. Our results suggest that in most cases, our method outperforms baseline methods when predicting motions. We also show that our method outperforms baselines for predicting human motion when a human and a robot share the workspace., Comment: 12 pages, Accepted for publication IEEE Transaction on Robotics 2016