skip to main content

Search for: All records

Creators/Authors contains: "Li, G."

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. Aims. An interplanetary coronal mass ejection (ICME) event was observed by the Solar Orbiter at 0.8 AU on 2020 April 19 and by Wind at 1 AU on 2020 April 20. Futhermore, an interplanetary shock wave was driven in front of the ICME. Here, we focus on the transmission of the magnetic fluctuations across the shock and we analyze the characteristic wave modes of solar wind turbulence in the vicinity of the shock observed by both spacecraft. Methods. The observed ICME event is characterized by a magnetic helicity-based technique. The ICME-driven shock normal was determined by magnetic coplanarity method formore »the Solar Orbiter and using a mixed plasma and field approach for Wind. The power spectra of magnetic field fluctuations were generated by applying both a fast Fourier transform and Morlet wavelet analysis. To understand the nature of waves observed near the shock, we used the normalized magnetic helicity as a diagnostic parameter. The wavelet-reconstructed magnetic field fluctuation hodograms were used to further study the polarization properties of waves. Results. We find that the ICME-driven shock observed by Solar Orbiter and Wind is a fast, forward oblique shock with a more perpendicular shock angle at the Wind position. After the shock crossing, the magnetic field fluctuation power increases. Most of the magnetic field fluctuation power resides in the transverse fluctuations. In the vicinity of the shock, both spacecraft observe right-hand polarized waves in the spacecraft frame. The upstream wave signatures fall within a relatively broad and low frequency band, which might be attributed to low frequency MHD waves excited by the streaming particles. For the downstream magnetic wave activity, we find oblique kinetic Alfvén waves with frequencies near the proton cyclotron frequency in the spacecraft frame. The frequency of the downstream waves increases by a factor of ∼7–10 due to the shock compression and the Doppler effect.« less
    Free, publicly-accessible full text available December 1, 2022
  2. DenseNets introduce concatenation-type skip connections that achieve state-of-the-art accuracy in several computer vision tasks. In this paper, we reveal that the topology of the concatenation-type skip connections is closely related to the gradient propagation which, in turn, enables a predictable behavior of DNNs’ test performance. To this end, we introduce a new metric called NN-Mass to quantify how effectively information flows through DNNs. Moreover, we empirically show that NN-Mass also works for other types of skip connections, e.g., for ResNets, Wide-ResNets (WRNs), and MobileNets, which contain addition-type skip connections (i.e., residuals or inverted residuals). As such, for both DenseNet-like CNNsmore »and ResNets/WRNs/MobileNets, our theoretically grounded NN-Mass can identify models with similar accuracy, despite having significantly different size/compute requirements. Detailed experiments on both synthetic and real datasets (e.g., MNIST, CIFAR-10, CIFAR-100, ImageNet) provide extensive evidence for our insights. Finally, the closed-form equation of our NN-Mass enables us to design significantly compressed DenseNets (for CIFAR-10) and MobileNets (for ImageNet) directly at initialization without time-consuming training and/or searching.« less
    Free, publicly-accessible full text available June 1, 2022
  3. Free, publicly-accessible full text available April 1, 2022
  4. This work contributes a generalized model for quantum computation called NChooseK. NChooseK is based on a single parametrized primitive suitable to express a variety of problems that cannot be solved efficiently using classical computers but may admit an efficient quantum solution. We implement a code generator that, given arbitrary parameters for N and K, generates code suitable for execution on IBM Q quantum hardware. We assess the performance of the code generator, limitations in the size of circuit depth and number of gates, and propose optimizations. We identify future work to improve efficiency and applicability of the NChooseK model.