Based on the quality and analysis of the medium carbon product whose grade was 81%-82%, the emphasis of the study was placed on the orthogonal experiments of roughing flotation reagent and single factors, such as abra...Based on the quality and analysis of the medium carbon product whose grade was 81%-82%, the emphasis of the study was placed on the orthogonal experiments of roughing flotation reagent and single factors, such as abrasive grain, pulp concentration and the rotating speed of flotation machine. The result with improvement in the grade of concentrate to 95% through the routine floatation method was achieved.展开更多
The state-of-art Computational Fluid Dynamics （CFD） codes FLUENT is applied in a fine-scale simulation of the wind field over a complex terrain. Several numerical tests are performed to validate the capability of FL...The state-of-art Computational Fluid Dynamics （CFD） codes FLUENT is applied in a fine-scale simulation of the wind field over a complex terrain. Several numerical tests are performed to validate the capability of FLUENT on describing the wind field details over a complex terrain. The results of the numerical tests show that FLUENT can simulate the wind field over extremely complex terrain, which cannot be simulated by mesoscale models. The reason why FLUENT can cope with extremely complex terrain, which can not be coped with by mesoscale models, relies on some particular techniques adopted by FLUENT, such as computer-aided design （CAD） technique, unstructured grid technique and finite volume method. Compared with mesoscale models, FLUENT can describe terrain in much more accurate details and can provide wind simulation results with higher resolution and more accuracy.展开更多
It is shown that the fine structure constant at Planck times tends to one as well as those of the weak and strong interactions. This results by constraining them at the Planck force. That seems to provide interesting ...It is shown that the fine structure constant at Planck times tends to one as well as those of the weak and strong interactions. This results by constraining them at the Planck force. That seems to provide interesting new results which confirm that at the beginning of space time (Planck scale) all fundamental forces converge to the same unit value.展开更多
To understand nutrient cycling in soils, soil processes and microorganisms need be better characterized. To determine whether specific trophic groups of fungi are associated with soil enzyme activity, we used soil imp...To understand nutrient cycling in soils, soil processes and microorganisms need be better characterized. To determine whether specific trophic groups of fungi are associated with soil enzyme activity, we used soil imprinting to guide mm-scale sampling from microsites with high and low phosphatase activities in birch/Douglas-fir stands. Study 1 involved sampling one root window per site at 12 sites of different ages(stands);study 2 was conducted at one of the stem-exclusion stands, at which 5 root windows had been installed. Total fungal and ectomycorrhizal(EM) fungal terminal-restriction fragment length polymorphism(TRFLP)fingerprints differed between high-and low-phosphatase activity microsites at 8 of 12 root windows across 12 sites. Where differences were detected, fewer EM fungi were detected in high-than low-phosphatase activity microsites. Using 5 root windows at one site,next-generation sequencing detected similar fungal communities across microsites, but the ratio of saprotrophic to EM fungal reads was higher in high-phosphatase activity microsites in the two windows that had low EM fungal richness. In windows with differences in fungal communities, both studies indicated that EM fungi were less successful than saprotrophic fungi in colonizing fine-scale,organic matter-rich microsites. Fine-scale sampling linked with in situ detection of enzyme activity revealed relationships between soil fungal communities and phosphatase activity that could not be observed at the scales employed by conventional approaches, thereby contributing to the understanding of fine-scale phosphorus cycling in forest soils.展开更多
This paper presents the development of a coarse-fine dual precision positioning stage to achieve long travel range and high accuracy.The fine stage is arranged in series with a coarse stage.The key in the fine stage d...This paper presents the development of a coarse-fine dual precision positioning stage to achieve long travel range and high accuracy.The fine stage is arranged in series with a coarse stage.The key in the fine stage design is the choice of a toggle mechanism for a tight mechanical loop with high stiffness and compactness.We designed the toggle mechanism for reduction of the displacement to suppress signal noises.The performance of the coarse and fine stages was verified with an optical encoder and capacitive sensor,respectively.The measurement results show that the dual mechanism has a travel range of 1 mm and resolution of 30 nm.展开更多
在细粒度图像分类中,现有的小样本学习算法未能充分结合通道和空间信息提取细粒度图像的判别性特征,导致仅依靠单一类型的特征不足以准确捕捉细粒度对象的类间差异.针对这一难题,提出了一种基于通道先验感知的多尺度细化网络,旨在有效...在细粒度图像分类中,现有的小样本学习算法未能充分结合通道和空间信息提取细粒度图像的判别性特征,导致仅依靠单一类型的特征不足以准确捕捉细粒度对象的类间差异.针对这一难题,提出了一种基于通道先验感知的多尺度细化网络,旨在有效融合通道信息和空间信息,提升小样本细粒度图像分类的性能.通道先验感知模块实现了通道维度上注意力权重的动态分配,从而高效地捕捉通道先验信息;多尺度特征聚合过程充分利用细粒度图像中丰富的上下文信息,获取丰富的空间和边界细节特征;最后,特征细化模块对上述提取的通道和空间维度信息进行优化,实现了对关键区域的动态选择和强调,进而融合形成更精细、更具代表性的混合特征表示.所提算法在以Conv-4作为骨干网络时,在Stanford Cars、Stanford Dogs和CUB-200-2011三个细粒度数据集上的实验分类性能显著提升.在5 way 1 shot分类任务中,三个数据集的准确率分别达到了79.95%、66.97%和81.91%;在5 way 5 shot分类任务中,准确率则分别为93.42%、82.48%和93.19%.展开更多
文摘Based on the quality and analysis of the medium carbon product whose grade was 81%-82%, the emphasis of the study was placed on the orthogonal experiments of roughing flotation reagent and single factors, such as abrasive grain, pulp concentration and the rotating speed of flotation machine. The result with improvement in the grade of concentrate to 95% through the routine floatation method was achieved.
基金supported by the National Natural Science Foundation of China(40805004, 40705039 and 90715031)the "Mini-projecton detailed survey and evaluation of wind energy resources"supported by National Climate Center of Chinese Meteoro-logical Administration (CWERA2010002)
文摘The state-of-art Computational Fluid Dynamics （CFD） codes FLUENT is applied in a fine-scale simulation of the wind field over a complex terrain. Several numerical tests are performed to validate the capability of FLUENT on describing the wind field details over a complex terrain. The results of the numerical tests show that FLUENT can simulate the wind field over extremely complex terrain, which cannot be simulated by mesoscale models. The reason why FLUENT can cope with extremely complex terrain, which can not be coped with by mesoscale models, relies on some particular techniques adopted by FLUENT, such as computer-aided design （CAD） technique, unstructured grid technique and finite volume method. Compared with mesoscale models, FLUENT can describe terrain in much more accurate details and can provide wind simulation results with higher resolution and more accuracy.
文摘It is shown that the fine structure constant at Planck times tends to one as well as those of the weak and strong interactions. This results by constraining them at the Planck force. That seems to provide interesting new results which confirm that at the beginning of space time (Planck scale) all fundamental forces converge to the same unit value.
基金supported by the Natural Sciences and Engineering Research Council of Canada (RGPIN170627-03, RGPIN170627-08 to M. D. Jones)the British Columbia Forest Innovation Investment—Forest Science Program (Y051064 to Suzanne Simard and M. D. Jones, Y081186 to M. D. Jones and S. J. Grayston)
文摘To understand nutrient cycling in soils, soil processes and microorganisms need be better characterized. To determine whether specific trophic groups of fungi are associated with soil enzyme activity, we used soil imprinting to guide mm-scale sampling from microsites with high and low phosphatase activities in birch/Douglas-fir stands. Study 1 involved sampling one root window per site at 12 sites of different ages(stands);study 2 was conducted at one of the stem-exclusion stands, at which 5 root windows had been installed. Total fungal and ectomycorrhizal(EM) fungal terminal-restriction fragment length polymorphism(TRFLP)fingerprints differed between high-and low-phosphatase activity microsites at 8 of 12 root windows across 12 sites. Where differences were detected, fewer EM fungi were detected in high-than low-phosphatase activity microsites. Using 5 root windows at one site,next-generation sequencing detected similar fungal communities across microsites, but the ratio of saprotrophic to EM fungal reads was higher in high-phosphatase activity microsites in the two windows that had low EM fungal richness. In windows with differences in fungal communities, both studies indicated that EM fungi were less successful than saprotrophic fungi in colonizing fine-scale,organic matter-rich microsites. Fine-scale sampling linked with in situ detection of enzyme activity revealed relationships between soil fungal communities and phosphatase activity that could not be observed at the scales employed by conventional approaches, thereby contributing to the understanding of fine-scale phosphorus cycling in forest soils.
基金supported by the National Science Council (No.NSC-96-2628-E002-199)
文摘This paper presents the development of a coarse-fine dual precision positioning stage to achieve long travel range and high accuracy.The fine stage is arranged in series with a coarse stage.The key in the fine stage design is the choice of a toggle mechanism for a tight mechanical loop with high stiffness and compactness.We designed the toggle mechanism for reduction of the displacement to suppress signal noises.The performance of the coarse and fine stages was verified with an optical encoder and capacitive sensor,respectively.The measurement results show that the dual mechanism has a travel range of 1 mm and resolution of 30 nm.
文摘在细粒度图像分类中,现有的小样本学习算法未能充分结合通道和空间信息提取细粒度图像的判别性特征,导致仅依靠单一类型的特征不足以准确捕捉细粒度对象的类间差异.针对这一难题,提出了一种基于通道先验感知的多尺度细化网络,旨在有效融合通道信息和空间信息,提升小样本细粒度图像分类的性能.通道先验感知模块实现了通道维度上注意力权重的动态分配,从而高效地捕捉通道先验信息;多尺度特征聚合过程充分利用细粒度图像中丰富的上下文信息,获取丰富的空间和边界细节特征;最后,特征细化模块对上述提取的通道和空间维度信息进行优化,实现了对关键区域的动态选择和强调,进而融合形成更精细、更具代表性的混合特征表示.所提算法在以Conv-4作为骨干网络时,在Stanford Cars、Stanford Dogs和CUB-200-2011三个细粒度数据集上的实验分类性能显著提升.在5 way 1 shot分类任务中,三个数据集的准确率分别达到了79.95%、66.97%和81.91%;在5 way 5 shot分类任务中,准确率则分别为93.42%、82.48%和93.19%.
