We explore the application of computer vision and machine learning (ML) techniques to predict material properties (e.g., compressive strength) based on SEM images of material microstructure. We show that it is possible to train ML models to predict materials performance based on SEM images alone, demonstrating this capability on the real-world problem of predicting uniaxially compressed peak stress of consolidated molecular solids (i.e. TATB) samples. Our image-based ML approach reduces root mean square error (RMSE) by an average of 51% over a non-image-based baseline. We compared two complementary approaches to this problem: (1) a traditional ML approach, random forest (RF), using state-ofthe-art computer vision features and (2) an end-to-end deep learning (DL) approach, where features are learned automatically from raw images. We demonstrate the complementarity of these approaches, showing that RF performs best in the "small data" regime in which many real-world scientific applications reside (up to 24% lower RMSE than DL), whereas DL outpaces RF in the "big data" regime, where abundant training samples are available (up to 24% lower RMSE than RF).
A 100 W amplified (75 W compressed) femtosecond (650 fs) Yb-fiber chirped-pulse-amplification system is demonstrated using broadband chirped-volume Bragg gratings (CVBGs) for the stretcher and compressor. With a 75% compression efficiency, the CVBG-based compressor exhibits an excellent average power handling capability and indicates the potential for further power scaling with this compact and robust technology.
The National Ignition Facility (NIF) at Lawrence Livermore National Laboratory, is the first of its kind megajoule-class laser facility with 192 beams capable of delivering over 1.8 MJ and 500TW of 351 nm light for high accuracy laser-matter interaction experiments. It has been commissioned and operated since 2009 to support a wide range of missions including the study of inertial confinement fusion, high energy density physics, material science, and laboratory astrophysics. In the first section of this paper we discuss the current status of laser performance obtained during the 408 target experiments completed in 2017. The performance spanned a wide range of laser energies, powers and pulse durations as requested for these target experiments. A special emphasis is given on energy delivery and cone power accuracy in the UV, as these are key parameters for successful experiments. In the second section of the paper, the results obtained during the 2017 performance quad campaign are briefly described. During this campaign a series of laser-only shots were taken to perform tests at elevated energies on a single NIF quad. These tests were designed to assess laser performance limits and operational costs against predictive models. This campaign culminated with the delivery of ~54 kJ of UV on a single quad of NIF, and 14 kJ on a single beam aperture, which are both to our knowledge the largest energies achieved to date for a neodymium-glass, frequency tripled architecture.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.