Levels of optimization

The concept of mass medication can hardly be accepted outside the agricultural field, but the experience of veterinary practitioners who treat large herds or flocks of animals fully supports this concept, especially in those areas of animal management where it is essential to keep all the animals at a level of optimal productivity (8). 

For one off separations flow rate optimization is not strictly necessary. However, it does offer an additional level of optimization for long-term projects and ongoing manufacture. The ideal flow rate often determined for small molecules on stationary phases with 10 [xm particle size and 60-300 A pores is approximately 120 cm/h, where for the majority of peptides and proteins the figure is typically in the region of 100 cm/h. 

For surface modification, to summarize the conclusions in Sect. 1, successful strategies usually consist of two levels of optimization optimal SMA design and optimal SMA localization. Combining the discussions in Sects. 1.2.3 and 1.3.2 about functional surface construction, the preference for SMA lo-... 

The formulation without operating variables is qualitatively easier to solve as one level of optimization is eliminated. More specifically, finding v for the optimization problem defined by Eq. (2) is a nondifferentiable global optimization problem that is extremely difficult to solve rigorously in the general case. It is therefore important to consider the pros and cons of using operating variables carefully. 

As shown in Fig. 3.7, keff is shown to increase significantly with lower fuel fractions. This is because the core is undermoderated, and increasing the carbon-to-uraniiun (C/U) ratio increases the moderation. For a pure carbon moderator with no poison, an overmoderated condition will never be attained. An increasing kes with lower fuel fraction does not mean that the bumup core lifetime is larger. In fact, there is typically a value for fuel fraction (for a given enrichment) where the bumup time is maximized. This level of optimization has not been studied yet, but it is expected that the optimum will occur near the value used for the standard configuration (i.e., around 0.3). 

Fig. 2. Examples of the dynamin I colorimetric GTPase assay. (A) L-a-phosphatidyl-L-serine (PS) liposome concentration curve. Stimulation of dynamin I GTPase activity was measured as a function of PS concentration using the colorimetric assay. The curve reaches a plateau at about 30 /xg/ml PS in this experiment, but varies between PS preparations. The curve is representative of the level of optimal dynamin activity in the absorbance range 0.4-0.6 OD units. (B) Concentration-dependent effect of compound A on dynamin I GTPase activity stimulated by 40 /xg/ml PS. The effect of the drug alone at increasing concentrations is shown in the open bars. The stimulation by PS + Compound A is shown in the hatched bars. Compound A reduces PS-stimulated dynamin I GTPase activity (hatched bars), until at high concentrations there appears to be stimulation. The solid bars show a subtraction of the other values, which, after eliminating the effect of background phosphate, reveals the full inhibitory curve of compound A.

The final product quality attributes are strongly related to the level of optimization of the composition of the liquid formulation, which is a multidisciplinary and challenging problem usually solved by tedious experimental approaches in the development of freeze-drying cycles. The topics of formulation that involve complex knowledge of physical chemistry, biochemistry and biology are outside the scope of this chapter. However, general concepts that are generally applied are summarized hereafter. 

Table 4.4 also shows the statistical data for the optimized circuits. In each case, the level of optimization effort selected (low, medium or high) made no difference to the circuit that was produced. 

All three levels of optimization effort were attempted - low, medium and high. In the first instance, only the component is optimized (preserving the hierarchy) The statistics for each stage are detailed in Table 4.8. In this table the values have been doubled to account for two identical components. The component s final circuit, generated after a high level of effort, is shown in Figure 4.38. It would appear that all redundant logic has been removed and even the delay on the critical path has been reduced. The optimization process has more than halved the area of each component. 

The patterns in the statistical data for both circuits are very similar. The biggest single reduction in area occurs during the low-effort level of optimization, when the circuit area is cut by half. The medium-effort level produces no improvement and the high level only a slight one. After all three optimization processes the circuits are identical apart from the presence of the top hierarchical level (Figure 4.35) in the first circuit. As this level contains only two component blocks it is purely cosmetic and does not affect circuit area or delays. 

Judging whether the circuit should be optimized as a complete block or as individual components is often hard. Although the total area of a flattened circuit may be smaller, the critical path may become complex and difficult to manage. If low, medium and dien high levels of optimization effort are applied the path may interconnect between several originally independent control functions and the relationships between component paths and fan-in or -out can become hard to imderstand. 

---