在MATLAB中了解3d矩阵

Question

我正在尝试在多元GARCH建模上使用一些MATLAB代码。

确切地说，我正在尝试建模

为此，我正在使用Kevin Shepards多元GARCH工具箱。

我不确定我最好使用什么程序（有很多），但是我继续使用matrix_garch，这听起来是最简单的。

此代码没有官方文档，但代码开头有一些信息...

我周围一团糟，我意识到如果我想使用参数选项，我需要插入6得到ap = 1，o = 0，q = 0和9（对于ap = 1，o = 0，q = 1。

是否有人可以告诉我是否甚至可以使用上图中的参数来模拟模型？所有k均设置为1，A矩阵为[0.2 0.7; 0 0.9]

非常感谢！

function [simulatedata,ht,pseudorc]=matrix_garch_simulate(t,k,parameters,p,o,q,m)
% Simulation of symmetric and asymmetric MATRIX multivariate GARCH models
%
% USAGE:
%   [SIMULATEDATA, HT, PSEUDORC] = matrix_garch_simulate(T, K, PARAMETERS, P, O, Q, M)
%
% INPUTS:
%   T            - Length of the time series to be simulated
%   K            - Cross-sectional dimension
%   PARAMETERS   - A 3-D matrix of K by K matrices where
%                    CC' = PARAMETERS(:,:,1), AA'(j)=PARAMETERS(:,:,1+j),
%                    GG'(j) = PARAMETERS(:,:,1+P+j), BB'(j)=PARAMETERS(:,:,1+P+Q+j)
%                    -OR- a K(K+1)/2*(1+P+O+Q) by 1 vector of parameters of
%                    the form returned my calling matrix_garch
%   P            - Positive, scalar integer representing the number of lags of the innovation process
%   O            - Non-negative scalar integer representing the number of asymmetric lags to include
%   Q            - Non-negative scalar integer representing the number of lags of conditional covariance
%   M            - [OPTIONAL] Number of ``intradaily'' returns to simulate to pseudo-Realized
%                    Covariance. If omitted, set to 72.
%
% OUTPUTS:
%   SIMULATEDATA - A time series with constant conditional correlation covariance
%   HT           - A [k k t] matrix of simulated conditional covariances
%   PSEUDORC     - A [k k t] matrix of pseudo-Realized Covariances
%
% COMMENTS:
%    The conditional variance, H(t), of a MATRIX GARCH is modeled as follows:
%
%      H(t) = CC' + AA'(1).*r_{t-1}'*r_{t-1} + ... + AA'(P).*r_{t-P}'*r_{t-P}
%                 + GG(1)'.*n_{t-1}'*n_{t-1} + ... + GG(O)'.*n_{t-P}'*n_{t-P}
%                  + BB(1)'.*H(t-1) +...+ BB(Q)'.*H(t-q)
%
%    where n_{t} = r_{t} .* (r_{t}<0).  If using realized measures, the
%    RM_{t-1} replaces r_{t-1}'*r_{t-1}, and the asymmetric version
%    replaces n_{t-1}'*n_{t-1}