Level hybrid models

As shown in Figure 9, the FPM model represents a generalization not only of DPD but also of the MD technique. It can be used as DPD by setting the noncentral forces to zero, or as MD by dropping the dissipative and Brownian components. These three techniques also can be combined into one three-level hybrid model. As shown in Figure 9, the three-level system consists of three different procedures representing each technique invoked in dependence on the type of particle interactions. We define three types of particles ... 

Prindle and Ray (ZB.) have recently analyzed the same styrene data using a hybrid model consisting of the micellar nucleation mechanism above the CMC and of the homogeneous nucleation and coagulation mechanism below the CMC. Their simulations show a much steeper rise in the particle number concentration precisely at the CMC than predicted by EPM. Their hybrid model does not appear to predict that the particle concentration levels off at high surfactant concentrations. 

An interesting alternative that combines the advantages of both classical and quantum mechanics is to use hybrid QM/MM models, first introduced by Arieh Warshel for modeling enzymatic reactions [7]. Here, the chemical species at the active site are treated using high-level (and therefore expensive) QM models, which are coupled to a force field that describes the reaction environment. Hybrid models can thus take into account solvent effects in homogeneous catalysis, support structure and interface effects in heterogeneous catalysis, and enzyme structure effects in biocatalysis. 

FIGURE 10.14 Hybrid model of the TS for hydrogenation of acetophenone by Ru(BINAP) (cyclohexanediamine)H2 (see Section 10.4.3). The central Ru(PH3)2(NH3)2H2 + CH20 system (shown in ball-and-stick format) is treated at a high level of theory, with the phenyl, naphthyl, methyl, and cyclohexyl groups (stick format) treated at a lower level. 

Increases in the molecular detail provided by modem analytical chemistry have motivated the formation of intermediate-level, molecular models. These hybrid models have molecules as reactants and products, not intermediate active centers (radicals or ions) or globally lumped product fractions. They embody more chemistry and rate constants... 

This increased the sulfate levels rather uniformly, but did not change the slope vs. distance up the Valley appreciably. The difficulty is that there are very few SO2 sources to the southwest of the west site closer than Texas. Possibly the observations could be fitted by including a more specific treatment of wet deposition. The lower frequency of rainfall in Texas relative to the Ohio River Valley may allow sulfur to survive transport from Texas to Kentucky more easily than transport up the Valley. Calculations using another type of hybrid model, that of Samson et al. (23,24), which can explicitly handle the removal by precipitation, might be the most successful approach to understanding these data. Because of their unique and comprehensive nature, these data are some of the most important to be fitted by models. No model should be considered reliable unless it can do so ... 

Hybrid models, of course, are even rarer than connectionist ones, despite numerous demands for them. A number of prominent researchers (from both sides of the symbolic/connectionist argument) have suggested that some union of the two representations is in order. For example, in The Computer and the Mind (1988), Philip Johnson-Laird hypothesized that one way to get around some of the dilemmas posed by existing models of cognition was to postulate different levels of representation high-level explicit symbols and low-level distributed symbolic patterns (p. 192). The opinion of a long-time connectionist is reflected in the title of a recent article Hybrid Computation in Cognitive Sci-... 

In this figure and others depicting hybrid models, we do not attempt to represent all units of the model. Rather, in the interest of having simple and easy-to-understand representations, we show only a few units at each level. Likewise, we do not show all rules that are part of the production system. 

Hendler, J. A. (1991). Developing hybrid symbolic/connectionist models. In J. A. Barnden 8c J. B. Pollack (Eds.), Advances in connectionist and neural computation theory High-level connectionist models, (Vol. 1, pp. 165-179). Norwood, NJ Ablex Publishing. 

Our treatment in this section will cover three primary thrusts in modeling dislocation core phenomena. Our first calculations will consider the simplest elastic models of dislocation dissociation. This will be followed by our first foray into mixed atomistic/continuum models in the form of the Peierls-Nabarro cohesive zone model. This hybrid model divides the dislocation into two parts, one of which is treated using linear elasticity and the other of which is considered in light of a continuum model of the atomic-level forces acting across the slip plane of the dislocation. Our analysis will finish with an assessment of the gains made in direct atomistic simulation of dislocation cores. 

Groppi et al. (D), (S) Different levels of model complexity to address simplified treatments in hybrid reactors... 

An assemble-to-order (ATO) system is a hybrid model of biuld-to-stock at the component (subassembly) level and assemble-to-order for the end product. In an ATO system, typically, the components take a substantial lead time to buUd, whereas the time it takes to assemble all the components into the final product is often negligible. Hence, keeping stock at the component level improves responsetime performance, whereas not keeping any end-product inventory reduces inventory cost and maximizes the flexibility for customization. A good example of an ATO system is the production of a PC (personal computer). Other examples include fast-food operations and many mail-order or e-commerce services. 

It must be reemphasized that hybridization is only a conceptual and a mathematical model that allows us to calculate molecular parameters. Changing hybridization is simply modifying that original model to suit a current need, just as the concept of h)4)ridization represents only a change to the model of atomic energy levels. Variable hybridization should not be considered more fundamental than the VSEPR model, just more mathematical. The ability to make quantitative predictions of molecular geometry and physical properties makes the variable hybridization model quite useful for some problems. On the other hand, the VSEPR model is also valuable as an intuitive basis for qualitatively correct predictions. As is so often the case, we need not decide which of two complementary models to adopt for all situations we need only to determine which best serves our purposes in a particular case. 

Schubert, J., Simutis, R., Dors, M., Havlik, I., and Liibbert, A. (1994) Hybrid modeling of yeast production processes -Combination of a priori knowledge on different levels of sophistication-. Chem. Eng. Technol. 17 10-20... 

---