In information theory, the typical set is a set of sequences whose probability is close to two raised to the negative power of the entropy of their source distribution. That this set has total probability close to one is a consequence of the asymptotic equipartition property (AEP) which is a kind of law of large numbers. The notion of typicality is only concerned with the probability of a sequence and not the actual sequence itself.
Contents
- Weakly typical sequences weak typicality entropy typicality
- Strongly typical sequences strong typicality letter typicality
- Jointly typical sequences
- Typical set encoding
- Typical set decoding
- References
This has great use in compression theory as it provides a theoretical means for compressing data, allowing us to represent any sequence Xn using nH(X) bits on average, and, hence, justifying the use of entropy as a measure of information from a source.
The AEP can also be proven for a large class of stationary ergodic processes, allowing typical set to be defined in more general cases.
(Weakly) typical sequences (weak typicality, entropy typicality)
If a sequence x1, ..., xn is drawn from an i.i.d. distribution X defined over a finite alphabet
Where
is the information entropy of X. The probability above need only be within a factor of 2nε.
It has the following properties if n is sufficiently large,
- The probability of a sequence from X being drawn from Aε(n) is greater than 1 − ε, i.e.
P r [ x ( n ) ∈ A ϵ ( n ) ] ≥ 1 − ϵ -
| A ε ( n ) | ⩽ 2 n ( H ( X ) + ε ) -
| A ε ( n ) | ⩾ ( 1 − ε ) 2 n ( H ( X ) − ε ) - Most sequences are not typical. If the distribution over
X is not uniform, then the fraction of sequences that are typical is
For a general stochastic process {X(t)} with AEP, the (weakly) typical set can be defined similarly with p(x1, x2, ..., xn) replaced by p(x0τ) (i.e. the probability of the sample limited to the time interval [0, τ]), n being the degree of freedom of the process in the time interval and H(X) being the entropy rate. If the process is continuous-valued, differential entropy is used instead.
Counter-intuitively, the most likely sequence is often not a member of the typical set. For example, suppose that X is an i.i.d Bernoulli random variable with p(0)=0.1 and p(1)=0.9. In n independent trials, since p(1)>p(0), the most likely sequence of outcome is the sequence of all 1's, (1,1,...,1). Here the entropy of X is H(X)=0.469, while
So this sequence is not in the typical set because its average logarithmic probability cannot come arbitrarily close to the entropy of the random variable X no matter how large we take the value of n. For Bernoulli random variables, the typical set consists of sequences with average numbers of 0s and 1s in n independent trials. For this example, if n=10, then the typical set consist of all sequences that has a single 0 in the entire sequence. In case p(0)=p(1)=0.5, then every possible binary sequences belong to the typical set.
Strongly typical sequences (strong typicality, letter typicality)
If a sequence x1, ..., xn is drawn from some specified joint distribution defined over a finite or an infinite alphabet
where
It can be shown that strongly typical sequences are also weakly typical (with a different constant ε), and hence the name. The two forms, however, are not equivalent. Strong typicality is often easier to work with in proving theorems for memoryless channels. However, as is apparent from the definition, this form of typicality is only defined for random variables having finite support.
Jointly typical sequences
Two sequences
Let
-
P [ ( X n , Y n ) ∈ A ε n ( X , Y ) ] ⩾ 1 − ϵ -
| A ε n ( X , Y ) | ⩽ 2 n ( H ( X , Y ) + ϵ ) -
| A ε n ( X , Y ) | ⩾ ( 1 − ϵ ) 2 n ( H ( X , Y ) − ϵ ) -
P [ ( X ~ n , Y ~ n ) ∈ A ε n ( X , Y ) ] ⩽ 2 − n ( I ( X ; Y ) − 3 ϵ ) -
P [ ( X ~ n , Y ~ n ) ∈ A ε n ( X , Y ) ] ⩾ ( 1 − ϵ ) 2 − n ( I ( X ; Y ) + 3 ϵ )
Typical set encoding
In information theory, typical set encoding encodes only the typical set of a stochastic source with fixed length block codes. Asymptotically, it is, by the AEP, lossless and achieves the minimum rate equal to the entropy rate of the source.
Typical set decoding
In information theory, typical set decoding is used in conjunction with random coding to estimate the transmitted message as the one with a codeword that is jointly ε-typical with the observation. i.e.
where