Heuristic methodologies

We presented a new heuristic methodology based on a Genetic Algorithm, for verifying the performance of a communication protocol in a realistic environment. We developed a prototypical implementation of the tool called Nepal and we ran some experiments, examining TCP connections in a simple IP network topology. 

Levitin et al. (1998) and Levitin (2004) extends the Universal Generating Function (UGF) technique of the multistate system analysis to the format when the performance distributions of some elements depend on the states of other elements. This useful new heuristic methodology may be used by researchers and some designers. 

The heuristic methodology we utilize is illustrated in Fig. 3. There are two kinds of resolution in the model, viz., the resolution of structure and dominant mechanisms. First, it is reasonable to resolve the flow structure into small bubbles, large bubbles, and hquid since there are plenty of reports in literature about the coexistence of small and large bubbles or a bimodal bubble size distribution (De Swart et al, 1996). This structure resolution is different from the spatial—temporal resolution of fluid domain in CFD models, and the significance fies in that the two bubble classes (TBCs) are actually relevant to the two kinds of dominant mechanisms. We will also show later that further resolution with more bubble classes in the model is not necessary and the MBSs are reduced to the TBCs. We use the following structure parameters for the description of the TBCs bubble diameters... 

The above reasoning is of a heuristic nature and is aimed at carrying out some analogy between the methods of the general theory of differential equations and those of the theory of difference schemes, the framework and methodology of which are outlined in this chapter. 

The program CHAOS has been mainly developed as a didactic tool i.e., as a heuristic aid for designing organic syntheses in such a manner that the student may use it, at his own pace, at different levels following the heuristic principles and methodologies developed in the present book. Two versions of CHAOS, one for IBM PCs (or "fully compatible" PCs) with Windows 3.1 or later, and another one for Macintosh computers are available. A summary of some of the main improvements of the present versions follows ... 

The other source on which our work draws is the heuristic research methodology. Heuristic programming involves trial and error procedures rather than algorithms, and has become more practical with the advent ot real-time interaction. The only condition for the practice... 

In this chapter we provide a heuristic introduction to scaling procedures that characterize properties of systems close to their critical point. The objective is to provide some general insights and to convey the flavor of the methodology. Much of the treatment of critical phenomena falls into the province of statistical mechanics, well outside the confines of classical thermodynamics. However, certain illuminating aspects that can be discussed without having to resort to statistical approaches will be taken up below. We center the discussion on critical exponents and the Landau theory of critical phenomena. For a proper exposition of the subject the reader is referred to several sources in the literature. ... 

Where sufficient data to enable a complete mathematical representation of the product-engineering problem exists, mathematical techniques exist for their solution. However, significant work is still needed to establish a comprehensive generic methodology to generate and systematically reduce the number of alternatives through heuristics, so that product engineering can be accomplished even in the absence of complete data. 

In chemical engineering, several methodologies and procedures are applied for decision making during design processes. Representative examples include design heuristics, shortcut methods, decision analysis methods, and mathematical optimization. 

The Decision Ontology should represent the different methodologies and procedures for evaluating and discriminating alternatives (e.g., heuristics, decision analysis methods) in a consistent way. 

However, there is a difference between conventional software algorithms and expert systems. Whereas conventional algorithms have a clearly defined result, an expert system may provide no answer or just one with a certain probability. The methodology used here is heuristic programming, and, depending on the point of view, the terms expert system and knowledge-based or rule-based system are often used synonymously. 

---