Useful multivariate distributions

Since the standard deviations are unity and the variables are independent (zero covariance), the covariance-matrix of X is the identity matrix / and the contours of constant probability in the space 9T are given by 

The surfaces of constant probability density are hyper-spheres. 

In the more general case, the vector Xn with mean p and nxn covariance matrix 

Contours of constant probability density in the space 91 are such as 

Parallel to the case of a single random variable, the mean vector and covariance matrix of random variables involved in a measurement are usually unknown, suggesting the use of their sampling distributions instead. Let us assume that x is a vector of n normally distributed variables with mean n-column vector ft and covariance matrix L. A sample of m observations has a mean vector x and annxn covariance matrix S. The properties of the t-distribution are extended to n variables by stating that the scalar m(x—p)TS ( —p) is distributed as the Hotelling s-T2 distribution. The matrix S/m is simply the covariance matrix of the estimate x. There is no need to tabulate the T2 distribution since the statistic 

---

Covariates are incorporated into the simulation as distributions that are either simulated stochastically or resampled from an existing database (18). Correlation between covariates is handled during stochastic simulations using multivariate distributions with appropriate variance-covariance structure. Alternatively, covariates resampled from a sufficiently large existing database carry all relevant covariates from an individual into a simulated individual and so capture inherent correlation. Regardless of the method, the simulated outputs for covariates need to be checked to ensure that they reflect the expected trial population and are consistent with trial inclusion and exclusion criteria. 

Assuming the distribution models are accurate and that they model all the possible behaviors in the data set, Bayes s theorem says that pup2, and p3 are the probabilities that the unknown sample is a member of class 1, 2, or 3, respectively. The distributions are modeled using multivariate Gaussian functions in a method known as expectation maximization. ... 

In a A -component mixture -1 independent variables are present. The multivariate distribution will then be formed using the variances of k- ... 

In SIMCA the distribution of the object in the inner model space is not considered, so the probability density in the inner space is constant and the overall PD appears as shown in Figs. 29, 30 for the enlarged and reduced SIMCA models. In CLASSY, Kernel estimation is used to compute the PD in the inner model space, whereas the errors in the outer space are considered, as in SIMCA, uncorrelated and with normal multivariate distribution, so that the overall distribution, in the inner and outer space of a one-dimensional model, looks like that reported in Fig. 31. Figures 32, 33 show the PD of the bivariate normal distribution and Kernel distribution (ALLOC) for the same data matrix as used for Fig. 31. Although in the data set of French wines no really important differences have been detected between SIMCA (enlarged model), ALLOC and CLASSY, it seems that CLASSY should be chosen when the number of objects is large and the distribution on the components of the inner model space is very different from a rectangular distribution. 

The most commonly employed univariate statistical methods are analysis of variance (ANOVA) and Student s r-test [8]. These methods are parametric, that is, they require that the populations studied be approximately normally distributed. Some non-parametric methods are also popular, as, f r example, Kruskal-Wallis ANOVA and Mann-Whitney s U-test [9]. A key feature of univariate statistical methods is that data are analysed one variable at a rime (OVAT). This means that any information contained in the relation between the variables is not included in the OVAT analysis. Univariate methods are the most commonly used methods, irrespective of the nature of the data. Thus, in a recent issue of the European Journal of Pharmacology (Vol. 137), 20 out of 23 research reports used multivariate measurement. However, all of them were analysed by univariate methods. 

H., Peck, C. C., Mould, D. R. Simulation of correlated continuous and categorical variables using a single multivariate distribution. J Pharmacokinet Pharmacodyn 2006 [Epub ahead of print]. 

Our understanding is that MPC has found widespread use in the petroleum industry. The chemical industry, however, is still dominated by the use of distributed control systems implementing simple PID controllers. We are addressing the plantwide control problem within this context. We are not addressing the application of multivariable model-based controllers in this book. 

This supervised classification method, which is the most used, accepts a normal multivariate distribution for the variables in each population ((Ai,..., A ) Xi) ), and calculates the classification functions minimising the possibility of incorrect classification of the observations of the training group (Bayesian type rule). If multivariate normality is accepted and equality of the k covariance matrices ((Ai,..., Xp) NCfti, X)), Linear Discriminant Analysis (LDA) calculates... 

Assuming a normal multivariate distribution, with the same covariance matrices, in each of the populations, (X, X2,..., Xp) V(7t , 5), the multivariate analysis of variance MANOVA) for a single factor with k levels (extension of the single factor ANOVA to the case of p variables), permits the equality of the k mean vectors in p variables to be tested Hq = jl = 7 2 = = where ft. = fl, fif,..., fVp) is the mean vector of p variables in population Wi. The statistic used in the comparison is the A of Wilks, the value of which can be estimated by another statistic with F-distribution. If the calculated value is greater than the tabulated value, the null hypothesis for equality of the k mean vectors must be rejected. To establish whether the variables can distinguish each pair of groups a statistic is used with the F-distribution with p and n — p — k + i df, based on the square of Mahalanobis distance between the centroids, that permits the equality of the pairs of mean vectors to be compared Hq = jti = ft j) (Aflfl and Azen 1979 Marti n-Alvarez 2000). 

The variable x in the preceding formulas denotes a quantity that varies. In our context, it signifies a reference value. If the variable by chance may take any one of a specified set of values, we use the term variate (i.e, a random variable). In this section, we consider distributions of single variates (i.e., univariate distributions). In a later section, we also discuss the joint distribution of two or more variates bivariate or multivariate distributions). 

In much the same way as the more common univariate statistics assume a normal distribution of the variable under study, so the most widely used multivariate models are based on the assumption of a multivariate normal distribution for each population sampled. The multivariate normal distribution is a generalization of its univariate counterpart and its equation in matrix notation is... 

In another extension for copolymerizations involving several monomers, Tobita has studied the full multivariate distribution of chain length and compositions for multi-component free radical polymerization by using generating functions [141, 142],... 

An early method developed uses multivariate normal distribution to model the classes on the basis of their data variances. Although this model has sometimes been used in analytical chemistry, it lacks general application because the method is based on the covariance matrix where it is tacitly assumed that many data exist and that the ratio between objects and variables is favorably about 6 1. 

Wajima T, Fukumura K, Yano Y, Oguma T. 2003. Prediction of human pharmacokinetics from animal data and molecular structural parameters using multivariate regression analysis volume of distribution at steady state. J Pham Phamacol 55 939-949. 

Continuous thermodynamics provides a simple way for the thermodynamic treatment of polydisperse systems. Such systems consist of a very large number of similar species whose composition is described not by the mole fractions of the individual components but by continuous distribution functions. For copolymers, multivariate distribution functions have to be used for describing the dependence of thermodynamic behavior on molar mass, chemical composition, sequence length, branching, etc. 

This review reports the state-of-art in the development and applications of continuous thermodynamics to copolymer systems characterized by multivariate distribution functions. Continuous thermodynamics permits the thermodynamic treatment of systems containing polydisperse homopolymers, polydisperse copolymers and other continuous mixtures by direct use of the continuous distribution functions as can be obtained experimentally. Thus, the total framework of chemical thermodynamics is converted to a new basis, the continuous one, and the crude method of pseudo-component splitting is avoided. 

The paper by Eide and Zahlsen also demonstrated the methodology for the identification and quantification of FAME in petrodiesel [12] by means of multivariate calibration (PLS regression). Dilution series were prepared with rapeseed or salmon biodiesel in petrodiesel in concentrations ranging from 0.5% to 10%.The spectrum of petrodiesel was shown to be visually different from the biodiesel spectra with a lower MW distribution. It is obviously beneficial for the identification and quantification of biodiesel in petrodiesel that the biodiesel and the petrodiesel have different but overlapping MW distributions however, it is not a prerequisite. Small differences between spectra can be detected using multivariate data analysis. 

For a DCL sensor of this kind, the information about the analyte is distributed over the entire spectrum. The spectrum therefore represents a fingerprint of the analyte. To correlate the spectral changes with the analyte properties of interest (identity, quantity, purity), it is advantageous to use multivariate analyses techniques. In this regard, a DCL sensor is related to sensor arrays (1, 2]. However, contrary to sensor arrays with independent sensor units, a DCL sensor is comprised of compounds that are connected by exchange reactions. Furthermore, the various sensors of an array have to be analyzed separately, whereas a single UV-Vis or fluorescence measurement is sufficient for a DCL sensor. 

The most important property for the characterization of particles is particle size. Randolph and Larson (36) pointed out that As no two particles will be exactly the same size, the material must be characterized by the distribution of sizes or particle-size distribution (PSD). If only size is of interest, a single-variable distribution function is sufficient to characterize the particulate system. If additional properties are also important, multivariable distribution functions must be developed. These distribution functions can be predicted through numerical simulations using population balance equations (PBE). 

Market simulations are run many times for a range of initial aquifer levels, with input obtained via Monte Carlo sampling from a joint, multivariate distribution created from inflow and withdrawal data. Simulated supply and demand conditions are translated into market prices for each transfer type, and the expected cost and reliability of various combinations, or portfolios, of transfer types can be computed. The transfer types are specified for each scenario, and a sequential search method is then used to identify minimum cost portfolios that meet designated supply-reliability constraints (Figure 3). Differences in the cost of the respective portfolios indicate the value of including each transaction type in the market, as well as how the cost of market-based approaches compares to the development of the least expensive new water source (Carrizo Aquifer). 

As easily inferable, the standard sampling approach cannot directly be employed. When multivariate distributions are used, RAVEN implements a surface search algorithm for identifying the iso-probability surface location. Once the location of the surface has been found, RAVEN chooses, randomly, one point on it. 

Scale- Up of Electrochemical Reactors. The intermediate scale of the pilot plant is frequendy used in the scale-up of an electrochemical reactor or process to full scale. Dimensional analysis (qv) has been used in chemical engineering scale-up to simplify and generalize a multivariant system, and may be appHed to electrochemical systems, but has shown limitations. It is best used in conjunction with mathematical models. Scale-up often involves seeking a few critical parameters. Eor electrochemical cells, these parameters are generally current distribution and cell resistance. The characteristics of electrolytic process scale-up have been described (63—65). 

---