Higher Dimensional Constructions

These examples are more complex than those described in Chapter 5. In this chapter, we also (finally) answer the BTX problem, which was originally posed in Chapter 1. 

---

This makes the physical body a kind of vessel which "contains" higher dimensional constructs. In three-dimensional terms it is logically absurd for a vessel to be smaller than its contents, yet this is not an inevitable conclusion if other dimensions are factored into the equation. Another way to conceptualize this is to think of space/time as the "outer" projection of an inner infinity. Imagine what it would be like to be a two-dimensional entity living on one of the six faces of a cube. 

This chapter marks the end of Section I of the book. All of the information discussed up to this point provides a firm foundation of the basics of AR theory. In the following chapters, we shall extend on these ideas, and relate them to higher dimensional constructions. Automated AR construction methods and variable density systems will also be discussed, which will allow us to tackle even more realistic problems. 

Whereas the profile in linear wave equations is usually arbitrary it is important to note that a nonlinear equation will normally describe a restricted class of profiles which ensure persistence of solitons as t — oo. Any theory of ordered structures starts from the assumption that there exist localized states of nonlinear fields and that these states are stable and robust. A one-dimensional soliton is an example of such a stable structure. Rather than identify elementary particles with simple wave packets, a much better assumption is therefore to regard them as solitons. Although no general formulations of stable two or higher dimensional soliton solutions in non-linear field models are known at present, the conceptual construct is sufficiently well founded to anticipate the future development of standing-wave soliton models of elementary particles. 

When the significant PC s have been extracted from X, the information left in the error matrix, E, can be used to estimate the residual variance of the model. This corresponds to constructing a tolerance volume around the PC model. This can be illustrated only with three-dimensional data and a one-dimensional PC model (Figure 6.5). There are no mathematical restrictions on estimating the residual variance in higher-dimensional models, but the human ability to visualize and understand spaces with more than three dimensions is limited. The residual variance can be calculated as... 

Support Vector Machine (SVM) is a classification and regression method developed by Vapnik.30 In support vector regression (SVR), the input variables are first mapped into a higher dimensional feature space by the use of a kernel function, and then a linear model is constructed in this feature space. The kernel functions often used in SVM include linear, polynomial, radial basis function (RBF), and sigmoid function. The generalization performance of SVM depends on the selection of several internal parameters of the algorithm (C and e), the type of kernel, and the parameters of the kernel.31... 

Local resolution can also be obtained by frequency variation, if a higher-dimensional electrochemical chain is constructed (analogous to the construction of a field effect transistor)298 such as... 

In the multidimensional case (i.e. N > 1), wave solutions are constructed in the phase space of the concentrations also called the hodograph space. For simplicity, the procedure will be demonstrated for a two-dimensional problem (i.e. N = 2). However, the same concepts apply to higher dimensional problems [16, 34, 38]. 

In this section we will develop the phase-space structure for a broad class of n-DOF Hamiltonian systems that are appropriate for the study of reaction dynamics through a rank-one saddle. For this class of systems we will show that on the energy surface there is always a higher-dimensional version of a saddle (an NHIM [22]) with codimension one (i.e., with dimensionality one less than the energy surface) stable and unstable manifolds. Within a region bounded by the stable and unstable manifolds of the NHIM, we can construct the TS, which is a dynamical surface of no return for the trajectories. Our approach is algorithmic in nature in the sense that we provide a series of steps that can be carried out to locate the NHIM, its stable and unstable manifolds, and the TS, as well as describe all possible trajectories near it. 

We therefore consider a different reaction flow model as our basic targeting model—one that can address temperature manipulation by feed mixing as well as by external heating or cooling. The model consists of a differential sidestream reactor (DSR), shown in Fig. 6, with a sidestream concentration set to the feed concentration and a general exit flow distribution function. (As mentioned in Section II, the boundary of an AR can be defined by DSRs for higher-dimensional (> 3) problems). We term this particular structure a cross-flow reactor. By construction, this model not only allows the manipulation of reactor temperature by feed mixing, but often eliminates the need to check for PFR extensions. 

As illustrated in Figure 5, SVMs work by constructing a hyper-plane in a higher-dimensional feature space of the input data,91 and use the hyper-plane (represented as Hi and H2) to enforce a linear separation of input samples, which belong to different classes (represented as Class O and Class X). The samples that lie on boundaries of different classes are referred to as support vectors. The underlying principle behind SVM-based classification is to maximize the margin between the support vectors using kernel functions. In the case of... 

Kernel methods, which include support vector machines and Gaussian processes, transform the data into a higher dimensional space, where it is possible to construct one or more hyperplanes for separation of classes or regression. These methods are more mathematically rigorous than neural networks and have in recent years been widely used in QSAR modeling. ... 

For any pD-model p 3), we can develop descriptors of variable dimensionality d. Examples of zero-dimensional descriptors are single numbers such as the radius of gyrationio (used for OD, ID, and 2D models) or the molecular volume 1 (used for 2D models). One-dimensional descriptors such as radial distribution functions or knot polynomials are used in OD and ID models, respectively. Two-dimensional descriptors include distance maps and Rama-chandran torsional-angle maps for some OD and ID models. Similarly, molecular graphs (2D descriptors) can be associated with ID models (contour lines), 2D models (molecular surfaces), or 3D models (e.g., the entire electron density function). Shape descriptors of higher dimensionality can also be constructed. 

The principle purpose of correlation experiments is to establish a one-to-one mapping from the signal to its source i.e. to the particular atomic nucleus in the molecule. This assignment task involves identification of the members in the coupling network, referred to as the spin system. In addition, correlation experiments, as such or with modifications, are suitable for measurements of scalar and dipolar couplings. Correlation in the two dimensions is the most natural dimensionality because the spin-spin interactions are pair wise. Three-dimensional or experiments of higher dimensionality are constructed from concatenated two-dimensional experiments. Homonuclear three-dimensional experiments, such as TOCSY-NOESY, are not considered here because in many cases the multidimensional heteronuclear experiments are superior. 

A mixed-valence complex of metal(dmit)2 (H2dmit = 4,5-dimercapto-l,3-dithiolo-2-thione) system is one of the promising candidates for the preparation of conductive LB films with higher dimensionality [27]. This sulfur-rich 1,2-dithiolene complex has provided three molecular superconductors, TTF[Ni(dmit)2]2 [28], TTF[Pd(dmit)2l2 [29], and Me4N[Ni(dmit)2]2 [30]. By the introduction of the alkylammonium group as a counter cation, the complexes become amphiphilic, and suitable for the construction of LB films. 

Furthermore, when AR constructions were carried out in previous chapters, they were described without adequate justification of why the resulting region represented the true AR. In order for these properties to be validated, and also in preparation for higher dimensional examples to be discussed (in Chapter 7), a more detailed AR theory, generalized to n-dimensional spaces, is required. This theory also assists in understanding of the kinds of reactor structures that should be expected when higher dimensional systems are considered in general. 

Readers interested in obtaining practical experience constructing higher dimensional ARs may choose to skip this chapter and move on to Chapter 7 instead, where a number of higher dimensional worked examples are provided. However, it will be useful to at least briefly read over the important results in this chapter before progressing on to Chapter 7. 

Although DSRs do not serve any important function in two-dimensional constructions, theorem 2 shows that they are a critical feature of the AR boundary in higher dimensions—critical DSRs serve as connectors to final PFR extreme points. Theorem 2 also describes the role that mixing and feed points play in relation to connectors on the boundary. Notice that if a connector satisfying theorem 2 is available, then we can conclude that mixing and feed points to connectors (critical CSTRs or DSRs) must take compositions from mixing lines (lineations) that also reside on the AR boundary. 

This discussion effectively completes the theory of attainable regions. (Attainable regions research is still relatively young, and many more results await discovery.) In the next chapter, we will look at applying many of the concepts and ideas discussed here to higher dimensional problems. These examples will mostly be for three-dimensional problems, so that the constructions are still easily visualized. 

ARs may also be generated for systems where a control parameter (often temperature) is employed. This theory is not part of the scope of this book, although we do provide a brief discussion of the relevant theory here for interest. The following discussion has ideas borrowed from Godorr et al. (1999), and all of the following discussions apply to alone. For higher dimensional problems, the use of an automated AR construction algorithm is often used instead. 

---