contents.m

多通道奇异谱分析程序

% MSSA - Tools for implementing Multichannel Singular Spectrum Analysis.
%-----------------------------------------------------------------------
% Written by Eric Breitenberger           Version date 1/11/96
%
% Please send comments and suggestions to [e-mail]
%-----------------------------------------------------------------------
% Files included in the toolkit:
%
% CONTENTS - This file.
%
% MSSA     - An example shell routine to call MSSAEIG, MPC, and MRC.
%            These routines are analogous to SSAEIG, PC, and RC for
%            single-channel SSA. 
%
% MSSAEIG  - Computes the data lag-covariance matrix and the T-EOFs. Calls 
%            COVAR, BK, AC, and XC. The covariance matrix can be generated
%            by using COVAR (FFT-based), AC and XC (direct summation), or
%            BK (Broomhead/King trajectory matrix based).
%
% MPC      - Calculates the principal components.
%
% MRC      - Calculates the reconstructed components.
%
% EOF      - Calculates the empirical orthogonal functions and principal
%            components of a dataset. In the context of MSSA, this function
%            is often used for data compression before an MSSA is performed.
%            For large data sets, use EOFCENT instead. 
%
% EOFCENT  - A function to calculate empirical orthogonal functions and 
%            principal components for a previously centered and normalized
%            data set. EOFCENT does the same thing as EOF, but does not allow
%            the data matrix to be modified within the function, thus 
%            avoiding the memory penalty of passing the large data matrix 
%            into the function.
%
% EEOF     - Computes Extended EOFs (e. g. Nasstrom and Weare 1982). This
%            procedure is mathematically equivalent to MSSA on a 
%            Broomhead-King type covariance matrix. The only differences are the
%            ordering of the output, and very small nemerical differences. EEOFs
%            are usually slower to compute than an MSSA, so the function is 
%            included here primarily as an illustrative example.
%
% AC       - Calculates autocovariance estimates for a time series.
%            N-weighted, N-k weighted, and Broomhead-King estimates are
%            available.
%
% XC       - Calculates cross-covariance estimates between two time series.
%            N-weighted, N-k weighted, and Broomhead-King estimates are
%            available.
%
% BK       - Calculates a Broomhead/King type covariance matrix for
%            a multichannel dataset. This covariance matrix is simply
%            the trajectory matrix multiplied by its transpose and
%            appropriately scaled.
%
% COVAR    - Calculates auto/cross-covariance estimates. In contrast to 
%            AC and XC which use direct summation, COVAR uses the FFT
%            to compute covariances. It is thus much faster for long series
%            and large lags. It only does N-weighted and N-k weighted
%            estimates - Broomhead-King type estimates are not amenable
%            to the FFT.
%
% POLARIZE - A simple utility to apply a uniform polarity to EOFs and PCs.
%            Often, it is desireable to compare the output from different
%            variants of MSSA. This can be complicated by the fact that the 
%            sign of EOFs and PCs is arbitrary. To eliminate this confusion,
%            POLARIZE applies the convention that the first element of each
%            EOF/PC should be positive.
%
% PLUCKRC  - A simple utility to pluck the i-th RC out of an RC matrix.
%