Multidimensional parameter space

More objective procedures in series design are clustering methods in multidimensional parameter space substituents from different clusters are selected for synthesis (chapter 3) [50, 154, 403]. As this approach cannot automatically avoid collinearity or multicollinearity, several different standard sets of aromatic substituents have been proposed e.g. [652, 653]). A distance mapping technique may be used to select further substituents on the basis of a nj ximum distance to the substituents which already are included [652]. A modification [654] of this approach uses the determinant of the parameter correlation matrix as the criterion for substituent selection. 

For series of compounds, the data obtained by several testing procedures constitute a multidimensional parameter space in which the inherent hidden... 

Cluster analysis is an exploratory data analysis technique aimed at grouping items (e.g. chemicals and their properties) into clusters of similar items according to their position in the multidimensional parameter space. The various clustering methods differ mainly in how they calculate the distance from a point to the cluster it may be to the nearest point, the most distant point or the centroid of the cluster, with the result that the respective clusters will have different shapes. 

In practice, often the periodic states need to be determined for optimization ends. This requires the computation of periodic states over a region in a multidimensional parameter space, with the aim to find an optimal p8irameter set. Especially in such a case the efficiency of the various methods to identify periodic states is of considerable importance. Thus, the ultimate test for these methods is to compare their performance in combination with optimization or continuation procedures. 

The first of these is that a least squares refinement will converge to the nearest minimum in the multidimensional parameter space, and unless the proposed starting model Is quite close to the correct structure this will not be the global minimum. Thus, the least squares approach will often converge to an incorrect structure. 

Chan EM et al (2010) Reproducible, high troughput synthesis of colloidal nanocrysttils for optimization in multidimensional parameter space. Nano Lett 10 1874-1885... 

From numerous results achieved using combinatorial and high-throughput methods, the most successful have been in the areas of molecular imprinting, polymeric compositions, catalytic metals for field-effect devices, and metal oxides for conductometric sensors. In those materials, the desired selectivity and sensitivity have been achieved by the exploration of multidimensional chemical composition and process parameters space at a previously unavailable level of detail at a fraction of time required for conventional one-at-a-time experiments. These new tools provided the opportunity for the more challenging, yet more rewarding explorations that previously were too time consuming to pursue. 

Hierarchical cluster analysis (HCA) and the closely related tree cluster analysis (TCA) provide a simple view of distances between samples, often viewed in a tree-like structure called a dendrogram (see Fig. 6a as an example). These types of analyses methods allow for the development of quick and simple classification schemes. Distances are calculated between all samples within the data set where the data parameters are the coordinates in a multidimensional variable parameter space (of dimension Mvar)- The general distance... 

Optimization implies that the method moves in a useful direction, at least eventually. In using a local gradient method, one can check this property by evaluating local function values and gradients, both of which should decrease monotonically, at least near the solution. However, during a search for a global minimum, the clues about the multidimensional landscape are more scattered, so it helps to run a number of searches in parallel. The GA is a prime example of a parallel optimization method that searches many regions of parameter space simultaneously. 

When I l is higher than Clf, by an amount equal to SU. then. v, = 1 w hcn If. is less than C l i by an equal amount,. v, = —1. and when I f = CT .. v, = 0. The center of interest xalues for all factors constitute the central point in the multidimensional factor space for the experiment. As indicated above, the constant /), is the value for r at this center in the factor space it is the (grand) average of all responses and is an important analysis output parameter for these designs. 

Parameter space is thus multidimensional so that we only discuss some illustrative cases. For a first set of values of the parameters Figure 3 describes the... 

If an assignment of classes to patterns is not evident, then unsupervised Learning methods are often helpful. Methods of finding clusters in a multidimensional pattern space are used to find natural classes in a data set. Such methods are not trivial and always contain heuristic and arbitrary elements. Subjective parameters are necessary to control the size, shape and number of clusters for a certain problem. different representations of the data often give different clusters. 

---