Control-flow hierarchy

A regular computation-intensive signal flow—as occurs in algebraic analysis, Altering, or format conversion— is combined with nested branches and multiple data-dependent loops. This explicit irregularity complicates not only the controller synthesis but also other synthesis tasks, Uke scheduling and data-path allocation. These tasks need to deal with control-flow hierarchy explicitly, which is an important aspect of our approach (see section 3). 

In this section, we further separate the sequencing graph hierarchy into two forms calling hierarchy and control-flow hierarchy. 

Figure 4.2 Example of the control-flow hierarchy for model M containing a loop vertex, which in turn contains a conditional vertex with two branches.

Hierarchy, This model uses hierarchy in two ways a calling hi chy that represents the nesting of procedure and function calls, and a control-flow hierarchy that represents conditional branching and loops. As a result, each graph in the model is acyclic. 

Synthesis across the hierarchy. As described in Chapter 4, the sequencing graph model supports two forms of hierarchy calling hierarchy, which refos to the nesting structure of model call vertices, and control-flow hierarchy, which refers to the nesting structure of conditionals and loops. 

A primary task in high-level architecture synthesis is the extraction of the data flow from a given algorithm description. The term data flow is defined here as the combination of operations and dependencies between them that define the algorithm. In contrast, control flow is then defined as a (partial) ordering of computations meeting the restrictions of their dependencies. The control flow is usually specified by introducing loops and function hierarchy. 

By analogy with the control loop hierarchy, one can split the control chain into the process to be controlled and the control system, as shown (for example) in Figure 12.18. In the given example, the process consists of the tank, which can be filled, the control valve with the appropriate piping, the restrictor, the level and flow transducers and perhaps a recorder the control system covers the programmable controller, the signal transmission, and the I/P converter. 

The logical deduction portion of the program is based on IF-THEN rules. FACTS, acquired both as the result of logical deductions and by querying the user, are stored in similar data structures. Because the branch points in the problem are also logical deductions, they are stored in a data structure similar to the FACTS. The branch points contain additional flow of control information that relates to the hierarchy of the problem. The difference between FACTS and branch points is transparent to the logical deduction portion of the program. 

The contributions of Vulpiani s group and of Kaneko deal with reactions at the macroscopic level. The contribution of Vulpiani s group discusses asymptotic analyses to macroscopic reactions involving flows, by presenting the mechanism of front formation in reactive systems. The contribution of Kaneko deals with the network of reactions within a cell, and it discusses the possibility of evolution and differentiation in terms of that network. In particular, he points out that molecules that exist only in small numbers can play the role of a switch in the network, and that these molecules control evolutionary processes of the network. This point demonstrates a limitation of the conventional statistical quantities such as density, which are obtained by coarse-graining microscopic quantities. In other words, new concepts will be required which go beyond the hierarchy in the levels of description such as micro and macro. 

Implementation hierarchy. Implement process control in a hierarchy based on frequency of decision making. The first level is protection (safety, environment, and equipment), and the second is smooth operation and stability (through control of flows, temperature, pressures, and levels and through alarms). The third level is product quality. The fourth level is protitabihty. The final level is monitoring and diagnosis. 

Fig. 7.6 Hierarchically clustered dendrogram calculated from the CMC analysis of the mechanism in fig. 7.1. The dendrogram is produced as described in step 7 in the text. The hierarchy is a good representation of the flow of control from the inputs down to Sg and S7. (From [1].)...

---