Generalized eigenvalue decomposition

Thus far we have considered the eigenvalue decomposition of a symmetric matrix which is of full rank, i.e. which is positive definite. In the more general case of a symmetric positive semi-definite pxp matrix A we will obtain r positive eigenvalues where r general case we obtain a pxr matrix of eigenvectors V such that ... 

Correspondence factor analysis can be described in three steps. First, one applies a transformation to the data which involves one of the three types of closure that have been described in the previous section. This step also defines two vectors of weight coefficients, one for each of the two dual spaces. The second step comprises a generalization of the usual singular value decomposition (SVD) or eigenvalue decomposition (EVD) to the case of weighted metrics. In the third and last step, one constructs a biplot for the geometrical representation of the rows and columns in a low-dimensional space of latent vectors. 

To solve for Ck, the generalized eigenvalue problem is used with the singular-value decomposition technique. The results of the problem indicate both the pure component response patterns x and y and the ratio of concentrations of the pure components to the standard response concentration. 

As in Section IV.A. the eigenvalues of the 1-RDM must lie in the interval [0,1] with the trace of each block equal to N/2. Similarly, with the a/a- and the jS/jS-blocks of the 2-RDM being equal, only one of these blocks requires purification. The purification of either block is the same as in Section IV.B.2 with the normalization being N N/2 — l)/4. The unitary decomposition ensures that the a/a-block of the 2-RDM contracts to the a-component of the 1-RDM. The purification of Section IV.B.2, however, cannot be directly applied to the a/jS-block of the 2-RDM since the spatial orbitals are not antisymmetric for example, the element with upper indices a, i fi, i is not necessarily zero. One possibility is to apply the purification to the entire 2-RDM. While this procedure ensures that the whole 2-RDM contracts correctly to the 1-RDM, it does not generally produce a 2-RDM whose individual spin blocks contract correctly. Usually the overall 1-RDM is correct only because the a/a-spin block has a contraction error that cancels with the contraction error from the a/ S-spin block. 

In its full generality, our third fundamental quantum-mechanical assumption says that the same kind of decomposition is possible with base states of the position observable, the momentum observable or indeed any observable. In other words, every observable has a complete set of base states. Typically the information about the base states and the value of the observable on each base state is collected into a mathematical object called a self-adjoint linear operator. The base states are the eigenvectors and the corresponding values of the observable are the eigenvalues. For more information about this point of view, see [RS, Section VIII.2]. 

Setting the ratio of the maximum deviatoric tensile component as Y and the isotropic tensile as Z, three eigenvalues are denoted as -Y/2 + Z, -Y/2 + Z and Y + Z. Then, it is assumed that the principal axes of the shear crack are identical to those of the tensile crack. As a result, the eigenvalues of the moment tensor for a general case are represented by the combination of the shear crack and the tensile crack. The following decomposition is obtained as the relative ratios X, Y and Z,... 

Fortunately, there are ways of determining the number of factors by looking at the igenvalues of the PCA factors. One fact that has been left out in all the discussions of PCA is that the data are not broken down into just two sets on values (scores and factors) but rather into three. The third set of values is the eigenvalues. Due to the way the PCA decomposition is calculated, the scores and factors generally span span a data range of + 1. If the scores and factors were the only representation of the data, all the principal component spectra would have the same relative intensities in the samples. Qearly, this is not the case some components will vary larger than others. 

---