Add distributions, tensor factories, and RNG operators to `Bonsai.ML.Torch`#79

Merged

glopesdev merged 16 commits into

bonsai-rx:mainfrom

ncguilbeault:dev/torch-distributions-random

Mar 12, 2026

Commits on Jan 13, 2026

Added Cdf and InverseCdf classes for probability distributions
ncguilbeault
committed
Added a `Generator` class to create a reproducible RNG with specified seed value
ncguilbeault
committed
Added a collection of probability distributions
ncguilbeault
committed
Added a class for seeding a torch compatible RNG
ncguilbeault
committed
Began adding a number of useful distributions and functionality to the distributions namespace
ncguilbeault
committed
Created a dedicated namespace for working with random numbers and added seed generator, as well as other random number factory methods
ncguilbeault
committed
Added overload to generator to support emitting generators based on elements arriving from input sequence
ncguilbeault
committed
Add support for using generator from input sequence and ignore in XML serialization
ncguilbeault
committed
Updated `Distributions` with doc strings and container base
ncguilbeault
committed
Added doc strings and made to inherit from tensor container base
ncguilbeault
committed
Refactored `Generator` documentation to suggest creating an individual RNG and added `ManualSeed` to set global state and return
ncguilbeault
committed
Renamed distributions for better consistency
ncguilbeault
committed
Refactored distribution classes with TensorOperatorConverter
ncguilbeault
committed
Refactored classes in the random namespace
ncguilbeault
committed
Updated XML documentation
ncguilbeault
committed

Commits on Mar 11, 2026

Renamed reparameterized sample to rsample for better alignment with torch
ncguilbeault
committed