Skip to content

Releases: leif-ibsen/SwiftRandom

Release 3.0.0

Choose a tag to compare

@leif-ibsen leif-ibsen released this 23 May 09:36

About SwiftRandom release 3.0.0:

  1. The program is reorganized so that it uses struct's instead of class'es.
    Among other things, this improves the performance considerably, especially for the
    Mersenne Twister generators.

  2. The MT32 Mersenne Twister has a jump function which advances the generator state
    as if 2^128 calls to the nextUInt32() method had been performed a number of times.

  3. Release 3.0.0 is backwards compatible with release 2.1.0 in the sense that release
    3.0.0 bitgenerators and release 2.1.0 bitgenerators will produce identical sequences
    of values if they are initialized with the same seed.

Release 2.1.0

Choose a tag to compare

@leif-ibsen leif-ibsen released this 09 May 14:16

About SwiftRandom release 2.1.0:

  1. The BitGenerator protocol inherits from the RandomNumberGenerator protocol
    so that the BitGenerator instances MT32, MT64, PCG32 and PCG64 can be used
    where a RandomNumberGenerator instance is required

  2. It is backwards compatible with release 2.0.0

Release 2.0.0

Choose a tag to compare

@leif-ibsen leif-ibsen released this 07 May 19:35

About SwiftRandom release 2.0.0 - changes from release 1.1.0:

  1. It generates random Int128 and UInt128 integers in addition to 64 bit and 32 bit integers

  2. Double values are no longer generated in closed ranges such as 1.0 ... 2.0,
    only in open ranges such as 1.0 ..< 2.0

  3. Three new methods to compute mean value, variance and standard deviation of a set of Double values

Release 1.1.0

Choose a tag to compare

@leif-ibsen leif-ibsen released this 30 Apr 08:22

About SwiftRandom release 1.1.0:

  1. Its API and functionality is unchanged from release 1.0.0

  2. A few minor documentation issues are fixed

Release 1.0.0

Choose a tag to compare

@leif-ibsen leif-ibsen released this 23 Oct 09:01

About SwiftRandom release 1.0.0:

  1. The SwiftRandom package generates various types of random numbers and distributions
    using either the Mersenne Twister algorithm or the Permuted Congruential Generator algorithm

  2. Some of this functionality is already present in the Digest package.
    It will eventually be removed from the Digest package in a new release.