This commit is contained in:
Miloslav Ciz 2023-04-02 20:57:34 +02:00
parent 138510fee5
commit 018a44958c
5 changed files with 112 additions and 4 deletions

View file

@ -8,7 +8,7 @@ Information is knowledge that can be used for making decisions. Information is i
In [computer science](compsci.md) the basic unit of information amount is 1 **[bit](bit.md)** (for *binary digit*), also known as [shannon](shannon.md). It represents a choice of two possible options, for example an answer to a *yes/no* question (with each answer being equally likely), or one of two [binary](binary.md) digits: 0 or 1. From this we derive higher units such as [bytes](byte.md) (8 bits), [kilobytes](memory_units.md) (1000 bytes) etc. Other units of information include [nat](nat.md) or [hart](hart.md). With enough bits we can encode any information including text, sounds and images. For this we invent various [formats](file_format.md) and encodings with different properties: some encodings may for example contain [redundancy](redundancy.md) to ensure the encoded information is preserved even if the data is partially lost. Some encodings may try to hide the contained information (see [encryption](encryption.md), [obfuscation](obfuscation.md), [steganography](steganography.md)). For processing information we create [algorithms](algorithm.md) which we usually execute with [computers](computer.md). We may store information (contained in data) in physical media such as [books](book.md), computer [memory](memory.md) or computer storage media such as [CDs](cd.md), or even with traditional potentially [analog](analog.md) media such as photographs.
Keep in mind that the **amount of physically present bits doesn't have to equal the amount of information** because, as mentioned above, data that takes *N* bits may e.g. utilize redundancy and so store less information that would theoretically be possible with *N* bits. It may happen that the stored bits are [correlated](correlation.md) for any reason or different binary values convey the same information (e.g. in some number encodings there are two values for number zero: positive and negative). All this means that the amount of information we receive in *N* bit data may be lower (but never higher) than *N* bits.
Keep in mind that the **amount of physically present bits doesn't have to equal the amount of information** because, as mentioned above, data that takes *N* bits may e.g. utilize redundancy and so store less information that would theoretically be possible with *N* bits. It may happen that the stored bits are [correlated](correlation.md) for any reason or different binary values convey the same information (e.g. in some number encodings there are two values for number zero: positive and negative). All this means that the amount of information we receive in *N* bit data may be lower (but never higher) than *N* bits, i.e. if we e.g. store a file on a 1 GB flash drive, the actual theoretical information contained may be lower -- the exact size of such theoretical information depends on probabilities of what can really appear in the file and MAY CHANGE with the knowledge we posses, i.e. the amount of information stored on the flash drive may change by simply us coming to know that the file stored on the drive is a movie about cats which rules out many combinations of bits that can be stored there. Imagine a simplified case when there is file which says whether there exists infinitely many [prime numbers](prime.md) -- to a mathematician who already knows the answer the file gives zero information, while to someone who doesn't know the answer the file provides 1 bit of information. However in practice we often make the simplification of equating the amount of physically present bits to the contained "information".
Information is related to **information [entropy](entropy.md)** (also Shannon entropy, similar to but distinct from the concept of thermodynamic entropy in physics); they're both measured in same units (usually [bits](bit.md)) but entropy measures a kind of "uncertainty" or average information received from a certain event when we know its probability distribution -- in a sense information and entropy can be seen as opposites: before we receive information we lack the information but there exists entropy, once we receive the information there is information but no entropy.