arXiv subjects

Yuhao Jiang

Publications and source records attributed to Yuhao Jiang.

26 records · Page 2Linked to original sources

arXiv2023

Neural MMO 2.0: A Massively Multi-task Addition to Massively Multi-agent Learning

Neural MMO 2.0 is a massively multi-agent environment for reinforcement learning research. The key feature of this new version is a flexible task system that allows users to define a broad range of objectives and reward signals. We challenge researchers to train agents capable of generalizing to tasks, maps, and opponents never seen during training. Neural MMO features procedurally generated maps with 128 agents in the standard setting and support for up to. Version 2.0 is a complete rewrite of its predecessor with three-fold improved performance and compatibility with CleanRL. We release the platform as free and open-source software with comprehensive documentation available at neuralmmo.github.io and an active community Discord. To spark initial research on this new platform, we are concurrently running a competition at NeurIPS 2023.

cs.AI↗

arXiv2023

Giant orbit-to-charge conversion induced via the inverse orbital Hall effect

We investigate the orbit-to-charge conversion in YIG/Pt/nonmagnetic material (NM) trilayer heterostructures. With the additional Ru layer on the top of YIG/Pt stacks, the charge current signal increases nearly an order of magnitude in both longitudinal spin Seebeck effect (SSE) and spin pumping (SP) measurements. Through thickness dependence studies of the Ru metal layer and theoretical model, we quantitatively clarify different contributions of the increased SSE signal that mainly comes from the inverse orbital Hall effect (IOHE) of Ru, and partially comes from the orbital sink effect in the Ru layer. A similar enhancement of SSE(SP) signals is also observed when Ru is replaced by other materials (Ta, W, and Cu), implying the universality of the IOHE in transition metals. Our findings not only suggest a more efficient generation of the charge current via the orbital angular moment channel but also provides crucial insights into the interplay among charge, spin, and orbit.

cond-mat.mtrl-sci↗

arXiv2023

Rethinking Class Activation Maps for Segmentation: Revealing Semantic Information in Shallow Layers by Reducing Noise

Class activation maps are widely used for explaining deep neural networks. Due to its ability to highlight regions of interest, it has evolved in recent years as a key step in weakly supervised learning. A major limitation to the performance of the class activation maps is the small spatial resolution of the feature maps in the last layer of the convolutional neural network. Therefore, we expect to generate high-resolution feature maps that result in high-quality semantic information. In this paper, we rethink the properties of semantic information in shallow feature maps. We find that the shallow feature maps still have fine-grained non-discriminative features while mixing considerable non-target noise. Furthermore, we propose a simple gradient-based denoising method to filter the noise by truncating the positive gradient. Our proposed scheme can be easily deployed in other CAM-related methods, facilitating these methods to obtain higher-quality class activation maps. We evaluate the proposed approach through a weakly-supervised semantic segmentation task, and a large number of experiments demonstrate the effectiveness of our approach.

cs.CV↗

arXiv2023

Orbitronics: Light-induced Orbit Currents in Terahertz Emission Experiments

Orbitronics is based on the use of orbit currents as information carriers. Up to now, orbit currents were created from the conversion of charge or spin currents, and inversely, they could be converted back to charge or spin currents. Here we demonstrate that orbit currents can also be generated by femtosecond light pulses on Ni. In multilayers associating Ni with oxides and nonmagnetic metals such as Cu, we detect the orbit currents by their conversion into charge currents and the resulting terahertz emission. We show that the orbit currents extraordinarily predominate the light-induced spin currents in Ni-based systems, whereas only spin currents can be detected with CoFeB-based systems. In addition, the analysis of the time delays of the terahertz pulses leads to relevant information on the velocity and propagation of orbit carriers. Our finding of light-induced orbit currents and our observation of their conversion into charge currents opens new avenues in orbitronics, including the development of orbitronic terahertz devices.

cond-mat.mes-hall↗

arXiv2023

Inverse Orbital Hall Effect Discovered from Light-Induced Terahertz Emission

Recent progress in orbitronics reveals the possibility of using orbit current as an information carrier. The interconversion between orbit currents and charge currents is crucial for orbit information processing. Although orbit currents can be created from charge currents via the orbital Hall effect, the conversion from orbit currents into charge currents has been observed only in very few systems due to the lack of a reliable orbit current source and the disturbance of the omnipresent inverse spin Hall effect. In this study, we show that ultrafast pulses of orbit current can be generated in Ni layers by femtosecond laser pulses. We demonstrate that, by injecting such orbit current pulses into nonmagnetic metals, a transient charge current is induced and emits terahertz electromagnetic pulses. The nonmagnetic metal layer acts as a converter of the orbit current into the charge current. The discovery of the generation and conversion of light-induced orbit current opens a new route for developing future orbitronic devices.

cond-mat.mes-hall↗

arXiv2022

Multi-Agent Path Finding via Tree LSTM

In recent years, Multi-Agent Path Finding (MAPF) has attracted attention from the fields of both Operations Research (OR) and Reinforcement Learning (RL). However, in the 2021 Flatland3 Challenge, a competition on MAPF, the best RL method scored only 27.9, far less than the best OR method. This paper proposes a new RL solution to Flatland3 Challenge, which scores 125.3, several times higher than the best RL solution before. We creatively apply a novel network architecture, TreeLSTM, to MAPF in our solution. Together with several other RL techniques, including reward shaping, multiple-phase training, and centralized control, our solution is comparable to the top 2-3 OR methods.

cs.AI↗

arXiv2022

Tunable Dynamic Walking via Soft Twisted Beam Vibration

We propose a novel mechanism that propagates vibration through soft twisted beams, taking advantage of dynamically-coupled anisotropic stiffness to simplify the actuation of walking robots. Using dynamic simulation and experimental approaches, we show that the coupled stiffness of twisted beams with terrain contact can be controlled to generate a variety of complex trajectories by changing the frequency of the input signal. This work reveals how ground contact influences the system's dynamic behavior, supporting the design of walking robots inspired by this phenomenon. We also show that the proposed twisted beam produces a tunable walking gait from a single vibrational input.

cs.RO↗

arXiv2022

Large deviations and fluctuation theorems for cycle currents defined in the loop-erased and spanning tree manners: a comparative study

The cycle current is a crucial quantity in stochastic thermodynamics. The absolute and net cycle currents of a Markovian system can be defined in the loop-erased (LE) or spanning tree (ST) manner. Here we make a comparative study between the large deviations and fluctuation theorems for LE and ST currents, i.e. cycle currents defined in the LE and ST manners. First, we derive the exact joint distribution and large deviation rate function for the LE currents of a system with a cyclic topology and also obtain the exact rate function for the ST currents of a general system. The relationship between the rate functions for LE and ST currents is clarified and the analytical results are applied to examine the fluctuations in the product rate of a three-step reversible enzyme reaction. Furthermore, we examine various types of fluctuation theorems satisfied by LE and ST currents and clarify their ranges of applicability. We show that both the absolute and net LE currents satisfy the strong form of all types of fluctuation theorems. In contrast, the absolute ST currents do not satisfy fluctuation theorems, while the net ST currents only satisfy the weak form of fluctuation theorems under the periodic boundary condition.

cond-mat.stat-mech↗