Search for: All records

An increasing magnetic field perpendicular to an undoped semiconductor surface at low temperature is known to strengthen the binding of localized electrons to stationary ions, as the wavefunction's tails evolve from exponential to Gaussian. It is also known that application of a high bias voltage to a depleted semiconductor can liberate bound charge and induce a large drop in electrical resistance. We connect these established results to experimental electrical transport measurements on off-state germanium Schottky-barrier metal–oxide–semiconductor field-effect transistor (MOSFETs) with an aluminum oxide insulating dielectric and platinum germanide contacts. We make measurements at the three distinct orientations of the magnetic field with respect to the substrate and the current. At 6 K, we observe sharp attenuation of current by more than 2 orders of magnitude, within 60 mT, at a crossover magnetic field perpendicular to the substrate. A 1 T magnetic field attenuates the current by more than 4 orders of magnitude. The strength of the attenuation and the value of the crossover field are controlled by both the gate–source and drain–source voltages. The attenuation is much weaker when the magnetic field is parallel to the current. Finally, we orient the magnetic field parallel to the substrate, but perpendicular to the current, allowing us to distinguish charge hopping at the oxide interface from charge hopping in the bulk. This large off-state magnetoresistance can be exploited for cryogenic magnetic- and photo-detection, and for high-bias, low-leakage MOSFETs. 

We present an algorithm for skill discovery from expert demonstrations. The algorithm first utilizes Large Language Models (LLMs) to propose an initial segmentation of the trajectories. Following that, a hierarchical variational inference framework incorporates the LLM-generated segmentation information to discover reusable skills by merging trajectory segments. To further control the trade-off between compression and reusability, we introduce a novel auxiliary objective based on the Minimum Description Length principle that helps guide this skill discovery process. Our results demonstrate that agents equipped with our method are able to discover skills that help accelerate learning and outperform baseline skill learning approaches on new long-horizon tasks in BabyAI, a grid world navigation environment, as well as ALFRED, a household simulation environment. 

We present a passive non-line-of-sight method that infers the number of people or activity of a person from the observation of a blank wall in an unknown room. Our technique analyzes complex imperceptible changes in indirect illumination in a video of the wall to reveal a signal that is correlated with motion in the hidden part of a scene. We use this signal to classify between zero, one, or two moving people, or the activity of a person in the hidden scene. We train two convolutional neural networks using data collected from 20 different scenes, and achieve an accuracy of 94% for both tasks in unseen test environments and real-time online settings. Unlike other passive non-line-of-sight methods, the technique does not rely on known occluders or controllable light sources, and generalizes to unknown rooms with no recalibration. We analyze the generalization and robustness of our method with both real and synthetic data, and study the effect of the scene parameters on the signal quality. 

Abstract Superfluid 3 He, with unconventional spin-triplet p-wave pairing, provides a model system for topological superconductors, which have attracted significant interest through potential applications in topologically protected quantum computing. In topological insulators and quantum Hall systems, the surface/edge states, arising from bulk-surface correspondence and the momentum space topology of the band structure, are robust. Here we demonstrate that in topological superfluids and superconductors the surface Andreev bound states, which depend on the momentum space topology of the emergent order parameter, are fragile with respect to the details of surface scattering. We confine superfluid 3 He within a cavity of height D comparable to the Cooper pair diameter ξ 0 . We precisely determine the superfluid transition temperature T c and the suppression of the superfluid energy gap, for different scattering conditions tuned in situ, and compare to the predictions of quasiclassical theory. We discover that surface magnetic scattering leads to unexpectedly large suppression of T c , corresponding to an increased density of low energy bound states. 

We recover a video of the motion taking place in a hidden scene by observing changes in indirect illumination in a nearby uncalibrated visible region. We solve this problem by factoring the observed video into a matrix product between the unknown hidden scene video and an unknown light transport matrix. This task is extremely ill-posed as any non-negative factorization will satisfy the data. Inspired by recent work on the Deep Image Prior, we parameterize the factor matrices using randomly initialized convolutional neural networks trained in a one-off manner, and show that this results in decompositions that reflect the true motion in the hidden scene.