The authors propose a neural network library N3LDG for natural language processing. N3LDG supports constructing computation graphs dynamically, and organizing executions into batches automatically. Experiments show that N3LDG can efficiently construct and execute computation graphs when training CNN, Bi-LSTM, and Tree-LSTM. When using CPU to train above models, the training speed of N3LDG is better than that of PyTorch. When using GPU to train CNN and Tree-LSTM, N3LDG is better than PyTorch.
In order to investigate soil seasonal gleization process in the alpine meadow on the Tibetan plateau, the soil gleization and its two main controls (high soil organic carbon content (SOC) content and anaerobic environment) were reported in the alpine grassland from Dec. 2013 to March 2014. The results show that the soil seasonal gleization occurres in the humus layer with blue-gray iron and manganese plaques. The soil seasonal gleization occures during the period of soil freezing and thawing, and a positive relationship between the thickness of gleization layer and frozen layer is observed. The process is pseudogley at mild level, with decreasing soil microbial biomass carbon, and increasing cation exchange capacity (CEC), active Fe2+, Mn2+ and availble phosphuros. The reults suggest that soil gleization may subsequently affect microbial community, root mortality, greenhouse gas emmission and nutrient cycling. Although soil gleization appears in the non-growing season, its impacts on the soil ecological processes should not be ignored.
The effect of nanoparticle or ionic silver on cadmium accumulation by lettuce (Lactuca sativa) was investigated. The lettuce plants were grown in Hoagland solution amended with 0, 10, 100, 1000 μg/L nanoparticle silver or 0.1, 1, 10 μg/L ionic silver. Half of the lettuce were exposed to 50 μg/L ionic cadmium. After 35 days of exposure, wet weight, dry weight, cadmium, chlorophyll and malondialdehyde of leaves and roots of the lettuces were measured. Results show that except for the 1000 μg/L nanoparticle silver which increased cadmium concentration by about 200% (p<0.001) in lettuce roots, other concentrations of nanoparticle or ionic silver do not significantly change the cadmium concentration in either leaves or roots. Biomass, chlorophyll and malondialdehyde of the lettuces are not significantly changed either. Nanoparticle silver applied as agricultural fungicide in practical amounts will not lead to increased cadmium accumulation or detectable phytotoxicity symptoms in lettuce.
This research takes oasis in the middle reaches of Heihe Basin in northwestern inland as study area, based on land use data, basic geographic information database, DEM, soil database, climatic data and DEMSP/OLS night light database, applies multiple time-series Logistic stepwise regression model to analyze driving factors for landscape types changes during 1986–1995, 1995–2000, 2000–2010, and reveals to what extent that water resource spatial distribution constrains to cropland and construction land. Conclusions are obtained as follows. 1) No general driving factors affect cropland increasing during each time period, while elevation and human activities present general driving factors for cropland decreasing. General driving factors for construction land increasing are elevation, distance and human activities, which have different driving directions based on different time periods. 2) Pumped wells and waterways share similar influential pattern on cropland and construction land changes with influential scope from 1 km to 1.5 km. Waterways constrain more than pumped wells. 3) The elevation influential scope on cropland and construction land is beneath 3500 m, with main constraint from topographic relief effect. The research results reveal the key instructional meanings for optimizing water and land resources spatial distribution, and improving utilization of water resource in oasis area.
The second generation larva of Argyrogramma albostriata of 4th instar, which were fed in the lab with Solidago canadensis under the condition of room temperature and natural light, were used as material for the experiment of host specify. 104 plant species from 52 families were used in testing host specify and the herbivory of leaves of different plants and survival time were observed. The results show that leaves of Conyza canadensis, Aster subulatus, Kalimeris indica, Erigeron annuus and S. canadensis were eaten by A. albostriata after 24 hours, while other plants were not eaten until the larva were starve to death. Then the life history of A. albostriata fed on C. canadensis, A. subulatus, K. indica, E. annuus and S. canadensis were observed and the main indexes of life history were recorded. The main results are as follows. 1) Survival rates of A. albostriata fed on C. canadensis, A. subulatus, K. indica, and S. canadensis are higher than larva fed on E. annuus significantly. 2) Fed on different plants have different influence on the length of different instars of A. albostriata. 3) Pupation rate, emergence rate, longevity and pupation weight of A. albostriata fed on C. canadensis, A. subulatus, K. indica, and S. canadensis are higher than that fed on E. annuus. The results indicate that A. albostriata is oligophagous, only eat C. canadensis, A. subulatus, K. isindica, E. annuus and S. canadensis. A. albostriata can complete its life history on C. canadensis, A. subulatus, K. indica, and S. Canadensis. However, the growth of lava of A. albostriata is constrained when fed on E. annuus.