=================================================================================

The flow is basically an underlying graph computation framework that uses tensors for its execution. A typical graph consists of two entities (as shown in Figure 4142a):
          i) Nodes. Nodes are also called vertices. The nodes are where actual computation (operation), such as addition, multiplication, division, etc., takes place (except for the leaf nodes).
          ii) Edges. The edges are actually the connections between the nodes/vertices through which the data flows.

Figure 4142a. Typical graph with nodes and edges.

The graph in Figure 4142a can normally be cyclic or acyclic. However, in TensorFlow, it is always acyclic, that is, it cannot start and end at the same node. As shown in Figure 4142b, the tensors flow through the edges or connections between nodes, and the computation at the next node results in formation of new tensors, such as d and e in the figure, is created, as an output constant or variable, through a computation at the node using other tensors a, b and c, and node Y, Y = (a + b) * (b-c), is created by using nodes (constant or variables) d and e. The nodes at the upper level become the input for next-node computation flowing through edges. The nodes at the same level can be executed in parallel, as there is no
interdependency between them. Therefore, d and e can be calculated in parallel at the same time. This attribute of graph helps to execute computational graphs in a distributed manner, which allows the TensorFlow to be used for large-scale applications.

============================================