skip to main content

Search for: All records

Creators/Authors contains: "Luo, Y."

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. Neural-network quantum molecular dynamics (NNQMD) simulations based on machine learning are revolutionizing atomistic simulations of materials by providing quantum-mechanical accuracy but orders-of-magnitude faster, illustrated by ACM Gordon Bell prize (2020) and finalist (2021). State-of-the-art (SOTA) NNQMD model founded on group theory featuring rotational equivari- ance and local descriptors has provided much higher accuracy and speed than those models, thus named Allegro (meaning fast). On massively parallel super- computers, however, it suffers a fidelity-scaling problem, where growing number of unphysical predictions of interatomic forces prohibits simulations involving larger numbers of atoms for longer times. Here, we solve this problem by com- bining the Allegro model with sharpness aware minimization (SAM) for enhanc- ing the robustness of model through improved smoothness of the loss landscape. The resulting Allegro-Legato (meaning fast and “smooth”) model was shown to elongate the time-to-failure tfailure, without sacrificing computational speed or accuracy. Specifically, Allegro-Legato exhibits much weaker dependence of time- to-failure on the problem size, t_failure = N^−0.14 (N is the number of atoms) compared to the SOTA Allegro model (t_failure ∝ N^−0.29), i.e., systematically delayed time-to-failure, thus allowing much larger and longer NNQMD simulations without failure. The model also exhibits excellent computational scalabil- ity and GPU acceleration on the Polaris supercomputer at Argonne Leadership Computing Facility. Such scalable, accurate, fast and robust NNQMD models will likely find broad applications in NNQMD simulations on emerging exaflop/s computers, with a specific example of accounting for nuclear quantum effects in the dynamics of ammonia to lay a foundation of the green ammonia technology for sustainability. 
    more » « less
    Free, publicly-accessible full text available May 10, 2024
  2. In this paper, we develop a novel procedure for low-rank tensor regression, namely Importance Sketching Low-rank Estimation for Tensors (ISLET). The central idea behind ISLET is importance sketching, i.e., carefully designed sketches based on both the responses and low-dimensional structure of the parameter of interest. We show that the proposed method is sharply minimax optimal in terms of the mean-squared error under low-rank Tucker assumptions and under the randomized Gaussian ensemble design. In addition, if a tensor is low-rank with group sparsity, our procedure also achieves minimax optimality. Further, we show through numerical study that ISLET achieves comparable or better mean-squared error performance to existing state-of-the-art methods while having substantial storage and run-time advantages including capabilities for parallel and distributed computing. In particular, our procedure performs reliable estimation with tensors of dimension $p = O(10^8)$ and is 1 or 2 orders of magnitude faster than baseline methods. 
    more » « less
  3. null (Ed.)
  4. A key goal of Next Generation Science Standards is to promote interest and exploration of natural phenomena. In preschool settings, teachers prompt exploration by asking questions, encouraging informal exploration and experimentation. To date, live or offline video observation has been the sole way to capture the quality of teacher question asking in the pre-k classroom (e.g., Sanders et al., 2016). To date, Automatic Speech Recognition (ASR) has not been used to measure the content/quality of teacher talk. Here, we used ASR to quantify preschool teachers’ use of keywords that promote student exploration and inquiry. 
    more » « less
  5. Abstract We present the Young Supernova Experiment Data Release 1 (YSE DR1), comprised of processed multicolor PanSTARRS1 griz and Zwicky Transient Facility (ZTF) gr photometry of 1975 transients with host–galaxy associations, redshifts, spectroscopic and/or photometric classifications, and additional data products from 2019 November 24 to 2021 December 20. YSE DR1 spans discoveries and observations from young and fast-rising supernovae (SNe) to transients that persist for over a year, with a redshift distribution reaching z ≈ 0.5. We present relative SN rates from YSE’s magnitude- and volume-limited surveys, which are consistent with previously published values within estimated uncertainties for untargeted surveys. We combine YSE and ZTF data, and create multisurvey SN simulations to train the ParSNIP and SuperRAENN photometric classification algorithms; when validating our ParSNIP classifier on 472 spectroscopically classified YSE DR1 SNe, we achieve 82% accuracy across three SN classes (SNe Ia, II, Ib/Ic) and 90% accuracy across two SN classes (SNe Ia, core-collapse SNe). Our classifier performs particularly well on SNe Ia, with high (>90%) individual completeness and purity, which will help build an anchor photometric SNe Ia sample for cosmology. We then use our photometric classifier to characterize our photometric sample of 1483 SNe, labeling 1048 (∼71%) SNe Ia, 339 (∼23%) SNe II, and 96 (∼6%) SNe Ib/Ic. YSE DR1 provides a training ground for building discovery, anomaly detection, and classification algorithms, performing cosmological analyses, understanding the nature of red and rare transients, exploring tidal disruption events and nuclear variability, and preparing for the forthcoming Vera C. Rubin Observatory Legacy Survey of Space and Time. 
    more » « less
    Free, publicly-accessible full text available May 1, 2024