KingOfNews (@kingofnews)

Just discovered this new Python project that has invented a new numeric representation that is superior to Rational Numbers and Floating Point Numbers called an “Arbitrary Number” which comes in two versions v1 and v2. V1 uses a looser definition where the number has an AST tree of fractions that can be combined using arbitrary operations while v2 is more conservative and doesn’t use an AST tree but still maintains exactness and infinite precision. According to the project, it is optimised for consumer Nvidia 32GB GPUs. It solves previously unsolved Mathematic problems in the domain of Machine Learning and Artificial Intelligence by retaining its exactness across all calculations and v1 outperforms other high performance operations since it can use optimised parallel deferred computation. Anyway it solves all the previous truncation issues with floating point numbers and Matrix Math inexactness and it is open source on GitHub so anyone can validate the unit tests and demonstrations and improve on it. The project is on GitHub under the Apache 2.0 licence at https://github.com/arbitrary-number/arbitrary-number .