1.      Markov sources

The state of the Markov chain is used to represent the “memory” of the source. The state could be the previous symbol from the source or could be the previous 300 symbols. It is possible to model as much memory as desired while staying in the regime of finite-state Markov chains.

(state與前M個symbols有關，就叫M重馬爾科夫鏈)

1. For example：

　　There is a binary source with outputs X1, X2,.... Assume that the symbol probabilities for Xk are conditioned on X(k-1) and X(k-2) but are independent 　　of all previous symbols given these past 2 symbols.

　　The alphabet of possible states is then the set of binary pairs, Sn = (Xn-1Xn), S = {[00], [01], [10], [11]}.

From the state S_k−1 = [01] (representing X_k−2=0,X_k−1=1), the output X_k=1 causes a transition to S_k = [11] (representing X_k−1=1,X_k=1). The chain is assumed to start at time 0 in a state S₀ given by some arbitrary pmf (probability mass function).

2.      Coding for Markov sources

3.      Lempel-Ziv universal data compression

(參考：https://www.jianshu.com/p/89dd96537d9d)

       兩種編碼壓縮方式：1，按概率去冗餘，如Huffman；2，去重複，如LZ77

       In general, two encoding compression methods: 1, compression by probability, such as Huffman algorithm; 2, deduplication, such as LZ77

&n