Search for: All records

Abstract Energetic particle deep penetration into low L‐shells (L < 4) impacts the dynamics of the radiation belts and ring current. Previous studies reported that electrons penetrate more frequently, deeply, and faster than protons of similar energies, but underlying mechanisms are unclear. In this study, we compare heavy‐ion behavior with electrons and protons to further identify the underlying mechanisms. Using Van Allen Probes data, we show that electron deep penetration occurs most frequently and deeply, followed by O⁺ions, then He⁺ions, and finally protons. Most particle deep penetrations occur within several hours. Superposed epoch analysis shows that prior to deep penetration, electrons have the steepest phase space density radial gradients, followed by heavy ions and then protons for the sameμandK. Our study suggests that a combination of two or more mechanisms, such as convection electric field and plasma wave‐induced scattering, may be needed to fully explain particle deep penetration. 

Battery-powered mobile devices (e.g., smartphones, AR/VR glasses, and various IoT devices) are increasingly being used for AI training due to their growing computational power and easy access to valuable, diverse, and real-time data. On-device training is highly energy-intensive, making accurate energy consumption estimation crucial for effective job scheduling and sustainable AI. However, the heterogeneity of devices and the complexity of models challenge the accuracy and generalizability of existing methods. This paper proposes AMPERE, a generic approach for energy consumption estimation in deep neural network (DNN) training. First, we examine the layer-wise energy additivity property of DNNs and strategically partition the entire model into layers for fine-grained energy consumption profiling. Then, we fit Gaussian Process (GP) models to learn from layer-wise energy consumption measurements and estimate a DNN's overall energy consumption based on its layer-wise energy additivity property. We conduct extensive experiments with various types of models across different real-world platforms. The results demonstrate that AMPERE has effectively reduced the Mean Absolute Percentage Error (MAPE) by up to 30%. Moreover, AMPERE is applied in guiding energy-aware pruning, successfully reducing energy consumption by 50%, thereby further demonstrating its generality and potential. 

We prepare quasi-1D films of Sb2Se3 on GaAs by molecular beam epitaxy. The aligned grains and anisotropic bonding hierarchy of the Sb2Se3 unit cell together produce giant birefringence in the near-infrared. 

Abstract The development of a deepening local minimum in phase space density (PSD)‐ profile indicates fast local loss potentially caused by wave‐induced scattering. The identification and characterization of proton PSD deepening minima are important for investigating the ring current loss and overall dynamics. Using multiyear Van Allen Probes observations, we analyze ∼10–100s keV proton PSD and report >100 keV proton deepening PSD minima for the first time. The overall occurrence rates of proton deepening local minimum peaks at ∼3%, mainly located at  = 4.5–5.0 near the plasmapause. The occurrence rate increases with the decrease of AL index and increase of solar wind dynamic pressure. The theoretical resonance energy of protons with typical He‐band electromagnetic ion cyclotron (EMIC) waves agrees with the energy of protons with deepening PSD minima. Thus, EMIC waves are the likely cause of the deepening PSD minimum and contribute to the fast local loss of ring current protons. 

Programmable photonic integrated circuits are expected to play an increasingly important role in enabling high-bandwidth optical interconnects and large-scale in-memory computing as needed to support the rise of artificial intelligence and machine learning technology. To that end, chalcogenide-based non-volatile phase-change materials (PCMs) present a promising solution due to zero static power. However, high switching voltage and a small number of operating levels present serious roadblocks to the widespread adoption of PCM-programmable units. Here, we demonstrate an electrically programmable wide bandgap Sb2S3-clad silicon ring resonator using a silicon microheater at a complementary-metal–oxide–semiconductor compatible voltage of <3 V. Our device shows a low switching energy of 35.33 nJ (0.48 mJ) for amorphization (crystallization) and reversible phase transitions with high endurance (>2000 switching events) near 1550 nm. Combining a volatile thermo-optic effect with non-volatile PCMs, we demonstrate 7-bit (127 levels) operation with excellent repeatability and reduced power consumption. Our demonstration of low-voltage and low-energy operation, combined with the hybrid volatile–nonvolatile approach, marks a significant step toward integrating PCM-based programmable units in large-scale optical interconnects. 

Abstract Programmable photonic integrated circuits (PICs) consisting of reconfigurable on-chip optical components have been creating new paradigms in various applications, such as integrated spectroscopy, multi-purpose microwave photonics, and optical information processing. Among many reconfiguration mechanisms, non-volatile chalcogenide phase-change materials (PCMs) exhibit a promising approach to the future very-large-scale programmable PICs, thanks to their zero static power and large optical index modulation, leading to extremely low energy consumption and ultra-compact footprints. However, the scalability of the current PCM-based programmable PICs is still limited since they are not directly off-the-shelf in commercial photonic foundries now. Here, we demonstrate a scalable platform harnessing the mature and reliable 300 mm silicon photonic fab, assisted by an in-house wide-bandgap PCM (Sb₂S₃) integration process. We show various non-volatile programmable devices, including micro-ring resonators, Mach-Zehnder interferometers and asymmetric directional couplers, with low loss (~0.0044 dB/µm), large phase shift (~0.012 π/µm) and high endurance (>5000 switching events with little performance degradation). Moreover, we showcase this platform’s capability of handling relatively complex structures such as multiple PIN diode heaters in devices, each independently controlling an Sb₂S₃segment. By reliably setting the Sb₂S₃segments to fully amorphous or crystalline state, we achieved deterministic multilevel operation. An asymmetric directional coupler with two unequal-length Sb₂S₃segments showed the capability of four-level switching, beyond cross-and-bar binary states. We further showed unbalanced Mach-Zehnder interferometers with equal-length and unequal-length Sb₂S₃segments, exhibiting reversible switching and a maximum of 5 ($$N+1,N=4$$ $N+1,N=4$) and 8 ($${2}^{N},N=3$$ $2^N,N=3$) equally spaced operation levels, respectively. This work lays the foundation for future programmable very-large-scale PICs with deterministic programmability.