Random
A module for obtaining randomness on the Internet Computer (IC).
This module provides the fundamentals for user abstractions to build on.
Dealing with randomness on a deterministic computing platform, such as the IC, is intricate. Some basic rules need to be followed by the user of this module to obtain (and maintain) the benefits of crypto- graphic randomness:
-
cryptographic entropy (randomness source) is only obtainable asyncronously in discrete chunks of 256 bits (32-byte sized `Blob`s)
-
all bets must be closed before entropy is being asked for in order to decide them
-
this implies that the same entropy (i.e.
Blob
) - or surplus entropy not utilised yet - cannot be used for a new round of bets without losing the cryptographic guarantees.
Concretely, the below class Finite
, as well as the
*From
methods risk the carrying-over of state from previous rounds.
These are provided for performance (and convenience) reasons, and need
special care when used. Similar caveats apply for user-defined (pseudo)
random number generators.
Finite
class Finite(entropy : Blob)
Drawing from a finite supply of entropy, Finite
provides
methods to obtain random values. When the entropy is used up,
null
is returned. Otherwise the outcomes' distributions are
stated for each method. The uniformity of outcomes is
guaranteed only when the supplied entropy is originally obtained
by the blob()
call, and is never reused.
byte
func byte() : ?Nat8
Uniformly distributes outcomes in the numeric range [0 .. 255]. Consumes 1 byte of entropy.
coin
func coin() : ?Bool
Simulates a coin toss. Both outcomes have equal probability. Consumes 1 bit of entropy (amortised).
byteFrom
func byteFrom(seed : Blob) : Nat8
Distributes outcomes in the numeric range [0 .. 255]. Seed blob must contain at least a byte.
coinFrom
func coinFrom(seed : Blob) : Bool
Simulates a coin toss. Seed blob must contain at least a byte.