tensorflow是什么意思 MXNet和Tensorflow的区别

tensorflow是什么意思

tensor flow
张量流
例句：
1
The resin flow behaviors in different gaps were studied by the unidirectional flow method together with the equivalent permeability tensor of flow path computed by two geometric models. In addition, the filling process was simulated to predict the influences of edge effects on the flow pattern.  
采用2种几何模型计算优先流道的等效渗透率张量，结合单向流动法研究不同缝隙宽度时树脂的流动行为，并进行充模过程仿真模拟，研究和预测边缘效应对充模流动的影响。
2
Tensor Analysis of Seepage Flow from Structural Plane Network of Rock Mass     
岩体结构面网格渗透张量分析方法

3
The simulation results show that the tensor method based power flow algorithm has better convergence, and can be used to solve ill-conditioned power flow problem.     
算例结果表明，基于张量方法的潮流算法具有更好的收敛性，可以用来求解病态潮流问题。

MXNet和Tensorflow的区别？

MXNet有两个主要的进程server和worker,worker之间不能进行通信,只能通过server互相影响。Tensorflow有worker,server,client三种进程,worker是可以相互通信的,可以根据op的依赖关系主动收发数据。MXNet常用来做数据并行,每个GPU设备上包含了计算图中所有的op,而Tensorflow可以由用户指定op的放置,一般情况下一个GPU设备负责某个和几个op的训练任务。