Engineering Papers⌕ Search

Engineering topics

Wang, Zheng

Publications and source records attributed to Wang, Zheng.

At least 19 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

OPER: Optimality-Guided Embedding Table Parallelization for Large-scale Recommendation Model

With the sharp increasing volume of user data, Deep Learning Recommendation Model (DLRM) becomes an indispensable infrastructure in large technology companies. However, large-scale DLRM on the multi-GPU platform is still inefficient due to unbalanced workload partitioning and intensive inter-GPU communication. To this end, we propose OPER, an OPtimality guided Embedding table placement for large-scale Recommendation model training and inference. OPER explores the potential of mitigating remote memory access latency in DLRM through fine-grained embedding table placement. Specifically, OPER proposes a theoretical modeling that builds up the relationship between EMT placement and the embedding communication latency in both training and inference. OPER proves the NP hardness of finding the optimal embedding table placement and proposes a heuristic algorithm that yields near optimal placement. OPER implements a SHMEM-based embedding table training system and a unified embedding index mapping to support fine-grained embedding table sharding and placement. Comprehensive experiments reveal that OPER achieves on average 3.4× and 5.1× speedup on training and inference respectively over state-of-the-art DLRM frameworks.

Wang, Zheng↗
DOE OSTI2024

RAP: Resource-aware Automated GPU Sharing for Multi-GPU Recommendation Model Training and Input Preprocessing

Ensuring high-quality recommendations for newly onboarded users requires the continuous retraining of Deep Learning Recommendation Models (DLRMs) with freshly generated data. To serve the online DLRM retraining, existing solutions use hundreds of CPU computing nodes designated for input preprocessing, causing significant power consumption that surpasses even the power usage of GPU trainers. To this end, we propose RAP, an end-to-end DLRM training framework that supports Resource-aware Automated GPU sharing for DLRM input Preprocessing and Training. The core idea of RAP is to accurately capture the remaining GPU computing resources during DLRM training for input preprocessing, achieving superior training efficiency without requiring additional resources. Specifically, RAP utilizes a co-running cost model to efficiently assess the costs of various input preprocessing operations, and it implements a resource-aware horizontal fusion technique that adaptively merges smaller kernels according to GPU availability, circumventing any interference with DLRM training. In addition, RAP leverages a heuristic searching algorithm that jointly optimizes both the input preprocessing graph mapping and the co-running schedule to maximize the end-to-end DLRM training throughput. The comprehensive evaluation shows that RAP achieves 78.3× speedup on average over CPU-based DLRM input preprocessing frameworks. In addition, the end-to-end training throughput of RAP is only 2.04% lower than the ideal case, which has no input preprocessing overhead.

Wang, Zheng↗
DOE OSTI2022

Variation in carbon and nitrogen concentrations among peatland categories at the global scale

Peatlands account for 15 to 30% of the world’s soil carbon (C) stock and are important controls over global nitrogen (N) cycles. However, C and N concentrations are known to vary among peatlands contributing to the uncertainty of global C inventories, but there are few global studies that relate peatland classification to peat chemistry. We analyzed 436 peat cores sampled in 24 countries across six continents and measured C, N, and organic matter (OM) content at three depths down to 70 cm. Sites were distinguished between northern (387) and tropical (49) peatlands and assigned to one of six distinct broadly recognized peatland categories that vary primarily along a pH gradient. Peat C and N concentrations, OM content, and C:N ratios differed significantly among peatland categories, but few differences in chemistry with depth were found within each category. Across all peatlands C and N concentrations in the 10–20 cm layer, were 440 ± 85.1 g kg -1 and 13.9 ± 7.4 g kg -1 , with an average C:N ratio of 30.1 ± 20.8. Among peatland categories, median C concentrations were highest in bogs, poor fens and tropical swamps (446–532 g kg -1 ) and lowest in intermediate and extremely rich fens (375–414 g kg -1 ). The C:OM ratio in peat was similar across most peatland categories, except in deeper samples from ombrotrophic tropical peat swamps that were higher than other peatlands categories. Peat N concentrations and C:N ratios varied approximately two-fold among peatland categories and N concentrations tended to be higher (and C:N lower) in intermediate fens compared with other peatland types. This study reports on a unique data set and demonstrates that differences in peat C and OM concentrations among broadly classified peatland categories are predictable, which can aid future studies that use land cover assessments to refine global peatland C and N stocks.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

Plutonium reactive transport in fractured granite: Multi-species experiments and simulations

Plutonium (Pu) in the subsurface environment can transport in different oxidation states as an aqueous solute or as colloidal particles. The transport behavior of Pu is affected by the relative abundances of these species and can be difficult to predict when they simultaneously exist. This study investigates the concurrent transport of Pu intrinsic colloids, Pu(IV) (aq) and Pu(V-VI) (aq) through a combination of controlled experiments and semi-analytical dual-porosity transport modeling. Pu transport experiments were conducted in a fractured granite at high and low flow rates to elucidate sorption processes and their scaling behavior. In the experiments, Pu(IV) (aq) was the least mobile of the Pu species, Pu(V-VI) (aq) had intermediate mobility, and the colloidal Pu, which consisted mainly of precipitated and/or hydrolyzed Pu(IV), was the most mobile. The semi-analytical modeling revealed that the sorption of each Pu species was rate-limited, as the sorption could not be described by assuming local equilibrium in the experiments. The model was able to describe the sorption of the different Pu species that occurring either on fracture surfaces, in the pores of the rock matrix, or simultaneously in both locations. While equally good fits to the data could be achieved using any of these assumptions, a fracture-dominated process was considered to be the most plausible because it provided the most reasonable estimates of sorption rate constants. Importantly, a key result of this work is that the sorption rate constant of all Pu species tends to decrease with increasing time scales, which implies that Pu will tend to be more mobile at longer time scales than observations at shorter time scales suggest. This result has important implications for predicting the environmental impacts of Pu in the safety assessments of geologic repositories for radioactive waste disposal, and we explore potential mechanistic bases for upscaling the sorption rate constants to time and distance scales that cannot be practically evaluated in experiments.

58 GEOSCIENCES↗
DOE OSTI2022

Latitude, Elevation, and Mean Annual Temperature Predict Peat Organic Matter Chemistry at a Global Scale

Peatlands contain a significant fraction of global soil carbon, but how these reservoirs will respond to the changing climate is still relatively unknown. A global picture of the variations in peat organic matter chemistry will aid our ability to gauge peatland soil response to climate. Here, the goal of this research is to test the hypotheses that (a) peat carbohydrate content, an indicator of soil organic matter reactivity, will increase with latitude and decrease with mean annual temperatures, (b) while peat aromatic content, an indicator of recalcitrance, will vary inversely, and (c) elevation will have a similar effect to latitude. We used Fourier Transform Infrared Spectroscopy to examine variations in the organic matter functional groups of 1034 peat samples collected from 10 to 20, 30–40, and 60–70 cm depths at 165 individual sites across a latitudinal gradient of 79°N–65°S and from elevations of 0–4,773 m. Carbohydrate contents of high latitude peat were significantly greater than peat originating near the equator, while aromatic content showed the opposite trend. For peat from similar latitudes but different elevations, the carbohydrate content was greater and aromatic content was lower at higher elevations. Higher carbohydrate content at higher latitudes indicates a greater potential for mineralization, whereas the chemical composition of low latitude peat is consistent with their apparent relative stability in the face of warmer temperatures. The combination of low carbohydrates and high aromatics at warmer locations near the equator suggests the mineralization of high latitude peat until reaching recalcitrance under a new temperature regime.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Atomic-Scale Imaging of Polarization Switching in an (Anti-)Ferroelectric Memory Material: Zirconia (ZrO2)

Direct, atomic-scale visualization of polarization switching in a functional, polycrystalline, binary oxide via in-situ high-resolution transmission electron microscopy (HRTEM) biasing is reported for the first time. Antiferroelectric (AFE) ZrO2 was used as the model system, which is important for commercial DRAMs and as emerging NVMs (through work-function engineering). We observed (1) clear shifting and coalescing of domains within a single grain, and (2) dramatic changes of the atomic arrangements and crystalline phases—both at voltages above the critical voltage measured for AFE switching. Similar synergistic in-situ structural-electrical characterization can pave the way to understand and engineer microscopic mechanisms for retention, fatigue, variability, sub-coercive switching and analog states in ferroelectric and AFE-based memory devices.

Lombardo, Sarah↗
DOE OSTI2019

Catalysts in Coronas: A Surface Spatial Confinement Strategy for High-Performance Catalysts in Methane Dry Reforming

One of the grand challenges in industrial catalytic processes is the inevitable sintering and aggregation of conventional supported catalysts to large particles, leading to the decrease of activity and even deactivation with time. Herein, a surface spatial confinement strategy was employed to design high-performing catalysts for the dry reforming of methane (DRM). Specifically, active nickel (Ni) nanoparticles (NPs) were confined on the surface of a dendritic mesoporous silica (DMS) in the form of the “catalysts in coronas”. The Ni/DMS catalyst exhibited a high catalytic performance close to its equilibrium conversion (76% conversion for CH4 at 700 °C). More importantly, the prepared catalyst remained stable after 145 h time-on-stream at 700 °C without noticeable carbon deposition. This sintering and coking resistance was found to arise from the surface spatial confinement effect in which the three-dimensional dendritic layers in the corona posted a steric barrier against migration and aggregation of Ni NPs and size of Ni NPs was controlled below 5 nm, hence against sintering and coking. Meanwhile, the mesoporous feature of the layered wall facilitated mass transport of reactants to Ni species and further boosted catalysis. This strategy should be broadly applicable to a range of metal- and metal oxide-supported catalysts in high-temperature heterogeneous reactions, such as DRM, water gas shift reaction, and vehicle emission control related reactions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2019

Active and stable Pt-Ceria nanowires@silica shell catalyst: Design, formation mechanism and total oxidation of CO and toluene

Cerium oxide is one of the most important rare earth metal oxides in catalysis, however, the sintering problem of noble metals and CeO 2 at higher temperature (e.g., >700 °C) is still unresolved. Herein, Pt nanoparticles self-assembled on ultra-thin CeO 2 nanowires (NWs) and then confined inside a thermally robust porous silica shell (Pt-CeO2NW@SiO 2 ) were introduced. The thickness of CeO 2 NWs was just ~2.0 nm. Moreover, Pt-CeO 2 NW@SiO 2 showed significantly enhanced catalytic performances for total oxidation of CO and toluene. The increased catalytic properties are attributed to the strong metal-support interaction effect between Pt and CeO 2 NWs at sub-nanoscale. Most importantly, the special core-shell structure also affords excellent sintering resistance retention up to 700 °C for 100 h, due to the guarding effect of porous silica shell. Finally, the formation mechanism of Pt-CeO 2 NW@SiO 2 was investigated in detail. Current strategy should inspire many rational designs of rare-earth metal-based nanocatalysts for real-world catalysts.

36 MATERIALS SCIENCE↗
DOE OSTI2018

Guided wave opto-acoustic device

The various technologies presented herein relate to various hybrid phononic-photonic waveguide structures that can exhibit nonlinear behavior associated with traveling-wave forward stimulated Brillouin scattering (forward-SBS). The various structures can simultaneously guide photons and phonons in a suspended membrane. By utilizing a suspended membrane, a substrate pathway can be eliminated for loss of phonons that suppresses SBS in conventional silicon-on-insulator (SOI) waveguides. Consequently, forward-SBS nonlinear susceptibilities are achievable at about 3000 times greater than achievable with a conventional waveguide system. Owing to the strong phonon-photon coupling achievable with the various embodiments, potential application for the various embodiments presented herein cover a range of radiofrequency (RF) and photonic signal processing applications. Further, the various embodiments presented herein are applicable to applications operating over a wide bandwidth, e.g. 100 MHz to 50 GHz or more.

Rakich, Peter Thomas↗
DOE OSTI2018

A comparison of mesoporous silica nanoparticles and mesoporous organosilica nanoparticles as drug vehicles for cancer therapy

Mesoporous silica nanoparticles (MSNs) are promising drug carriers for use in cancer treatment owing to their excellent biocompatibility and drug‐loading capacity. However, MSN's incomplete drug release and toxic bioaccumulation phenomena limit their clinical application. Recently, researchers have presented redox responsive mesoporous organosilica nanoparticles containing disulfide (S–S) bridges (ss‐MONs). These nanoparticles retained their ability to undergo structural degradation and increased their local release activity when exposed to reducing agents. Disulfide‐based mesoporous organosilica nanoparticles offer researchers a better option for loading chemotherapeutic drugs due to their effective biodegradability through the reduction of glutathione. Although the potential of ss‐MONs in cancer theranostics has been studied, few researchers have systematically compared ss‐MONs with MSNs with regard to endocytosis, drug release, cytotoxicity, and therapeutic effect. In this work, ss‐MONs and MSNs with equal morphology and size were designed and used to payload doxorubicin hydrochloride (DOX) for liver cancer chemotherapy. The ss‐MONs showed considerable degradability in the presence of glutathione and performed comparably to MSNs on biocompatibility measures, including cytotoxicity and endocytosis, as well as in drug‐loading capacity. Notably, DOX‐loaded ss‐MONs exhibited higher intracellular drug release in cancer cells and better anticancer effects in comparison with DOX‐loaded MSNs. Hence, the ss‐MONs may be more desirable carriers for a highly efficient and safe treatment of cancer.

Yue, Juan↗
DOE OSTI2018

Interferometric imaging of nonlocal electromechanical power transduction in ferroelectric domains

Significance The conversion between electrical and acoustic signals in piezoelectric materials is of fundamental importance for their applications. Because of the much slower speed of sound than speed of light in solids, mesoscopic imaging is imperative for the study of electroacoustic behaviors at microwave frequencies. In this paper, the electromechanical power transduction in lithium niobate domains is spatially resolved by microwave impedance microscopy. Because of the sign reversal of piezoelectric tensor in opposite domains, the interaction between electric fields and elastic waves leads to fringe patterns that are fundamentally different from the acoustic displacement fields. This approach uncovers hidden information in the piezoelectric transduction process and opens a frontier to explore various elastic phenomena in materials and devices.

42 ENGINEERING↗
DOE OSTI2018

The Flash ADC system and PMT waveform reconstruction for the Daya Bay experiment

To better understand the energy response of the Antineutrino Detector (AD), the Daya Bay Reactor Neutrino Experiment installed a full Flash ADC readout system on one AD that allowed for simultaneous data taking with the current readout system. This paper presents the design, data acquisition, and simulation of the Flash ADC system, and focuses on the PMT waveform reconstruction algorithms. For liquid scintillator calorimetry, the most critical requirement to waveform reconstruction is linearity. Several common reconstruction methods were tested but the linearity performance was not satisfactory. A new method based on the deconvolution technique was developed with 1% residual non-linearity, which fulfills the requirement. Here, the performance was validated with both data and Monte Carlo (MC) simulations, and 1% consistency between them has been achieved.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2017

Observation of Third-order Nonlinearities in Graphene Oxide Film at Telecommunication Wavelengths

All-optical switches have been considered as a promising solution to overcome the fundamental speed limit of the current electronic switches. However, the lack of a suitable third-order nonlinear material greatly hinders the development of this technology. Here in this paper we report the observation of ultrahigh third-order nonlinearity about 0.45 cm 2 /GW in graphene oxide thin films at the telecommunication wavelength region, which is four orders of magnitude higher than that of single crystalline silicon. Besides, graphene oxide is water soluble and thus easy to process due to the existence of oxygen containing groups. These unique properties can potentially significantly advance the performance of alloptical switches.

36 MATERIALS SCIENCE↗
DOE OSTI2017

Development of an integrated four-channel fast avalanche-photodiode detector system with nanosecond time resolution

In this paper, a four-channel nanosecond time-resolved avalanche-photodiode (APD) detector system is developed at Beijing Synchrotron Radiation. It uses a single module for signal processing and readout. This integrated system provides better reliability and flexibility for custom improvement. The detector system consists of three parts: (i) four APD sensors, (ii) four fast preamplifiers and (iii) a time-digital-converter (TDC) readout electronics. The C30703FH silicon APD chips fabricated by Excelitas are used as the sensors of the detectors. It has an effective light-sensitive area of 10 × 10 mm 2 and an absorption layer thickness of 110 μ m. A fast preamplifier with a gain of 59 dB and bandwidth of 2 GHz is designed to readout of the weak signal from the C30703FH APD. The TDC is realized by a Spartan-6 field-programmable-gate-array (FPGA) with multiphase method in a resolution of 1ns. The arrival time of all scattering events between two start triggers can be recorded by the TDC. The detector has been used for nuclear resonant scattering study at both Advanced Photon Source and also at Beijing Synchrotron Radiation Facility. Finally, for the X-ray energy of 14.4 keV, the time resolution, the full width of half maximum (FWHM) of the detector (APD sensor + fast amplifier) is 0.86 ns, and the whole detector system (APD sensors + fast amplifiers + TDC readout electronics) achieves a time resolution of 1.4 ns.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2017

On-chip photonic-phononic emitter-receiver apparatus

A radio-frequency photonic devices employs photon-phonon coupling for information transfer. The device includes a membrane in which a two-dimensionally periodic phononic crystal (PnC) structure is patterned. The device also includes at least a first optical waveguide embedded in the membrane. At least a first line-defect region interrupts the PnC structure. The first optical waveguide is embedded within the line-defect region.

47 OTHER INSTRUMENTATION↗