hmm-gmm implementation

Hello,

I am a beginner in MATLAB and HMM, and I have to implement a continuous hmm. After research i found the Kevin Murphy's code (<http://http://www.cs.ubc.ca/~murphyk/Software/HMM/hmm_usage.html>) that implements HMMs with mixture of Gaussians outputs, but I can not understand some poits.

1)

Let us generate nex=50 vector-valued sequences of length T=50; each vector has size O=2.
O = 2;
T = 50; 
nex = 50; 
data = randn(O,T,nex);

For this part of the code I can not understand what is O. If I understood correctly it generates randomly 50 data vectors (nex) of size 50 (T), but O = 2, I do not understand its meaning.

2)

[mu0, Sigma0] = mixgauss_init(Q*M, reshape(data, [O T*nex]), cov_type);
mu0 = reshape(mu0, [O Q M]);
Sigma0 = reshape(Sigma0, [O O Q M]);
mixmat0 = mk_stochastic(rand(Q,M));

The last line I suppose it randomly generates the covariance matrix but the first three lines I can not understand.

3)

[LL, prior1, transmat1, mu1, Sigma1, mixmat1] = ...
mhmm_em(data, prior0, transmat0, mu0, Sigma0, mixmat0, 'max_iter', 2);

There normally is the line for learning but I do not understand what is 'max_iter', 2 and LL.

Thank you.