entropy - calculate data entropy
This manual describes the operating procedure for the entropy
This program will calculate the specific entropy for a given chunk of binary
or ascii data that was previously read from stdin
or supplied on the
command line via -t
The most common use for this program is to gauge the efficiency of a compression
algorithm. In theory, based on the data's entropy, the maximum theoretical
compression rate can be calculated.
To understand how compression works, we must understand that all data can be
characterized by informational content, called its entropy
(the term is
borrowed from Thermo-Dynamics). Compression is possible because most data is
represented with more bits then its entropy suggests is optimal.
- Rather then read the data from the stdin input stream, use the
string string as the source of the data. This is useful when you
would like to calculate for a specific word or sentence.
- Sorted output based on symbol frequencies
- Print entropy version.
A data set's entropy is the sum of each distinct symbol's entropies. A symbol is
a distinct set of contiguous bits. For the sake of simplicity, I use 8 (number
of bits making up a byte). It is important to note that a symbol does
have to consist of 8 bits. A distinct symbols entropy `S' of symbol
'z' can be defined as:
Sz = -log2(Pz)
Where `Pz' is the probability
of symbol 'z' being found in the set. By
storing each distinct symbol as a node in a list, we can calculate exactly how
many times the symbol occurs in the set. This is also known as solving for the
In order to calculate the overall entropy of the data; we sum the total
entropies contributed by each distinct symbol and divide that by a coefficient
known as the data size
. This coefficient is calculated by taking the
number of bits per symbol and multiplying it by the total quantity of symbols.
The entropy utility exits 0 on success, and non-zero if an error occurs.
gzip(1), bzip2(1), tar(1), minigzip(1), gzexe(1), kgzip(8)
compress(1), zcmp(1), zdiff(1), zlib(3)
o Chris S.J. Peron
None known. This does not mean they do not exist though. Please send bug reports
and source code patches to (firstname.lastname@example.org).