hal_0013

We are announcing our newest initiative: an interactive math engine by @isaac_mason_. Math is core to everything interactive from geometry to collision to color, yet JavaScript lacks a high-performance kernel. Instead libraries reinvent math structures and transformations, with varying success, and now LLMs generate bespoke functions on a case by case basis. The problem is that there are no principled guarantees for correctness or performance leading to degraded apps or difficult to read code. We need math engine that can be used to power all the cool projects people want to make at scale! The goals of the initiative are ambitious: ⟡ Predictable performance. Allocation-free operations and a documented usage contract designed to preserve monomorphic, optimizable call sites, validated through reproducible benchmarks. ⟡ Portable. Efficiently interoperates with WebGL, WebGPU, Wasm, Three.js, etc., so that the handoff between math kernel and framework is simple. ⟡ Minimal. A lean, tree-shakable kernel containing only the primitives needed to build interactive algorithms. ⟡ Data-oriented. Operates on caller-owned data through data-in, data-out functions without owning the data lifecycle. ⟡ Readability. It is important the core math operations readable such that even if the code is not written by a person they can reasonably review it. @isaac_mason_ is bringing his experience building out navcat, crashcat and gpucat. We are starting with mathcat as the foundation and building from there. Stay tuned for more. 👉 Initiative: github.com/pmndrs/org/blob/m… 👉 Join the convo on Discord: discord.com/invite/poimandre…