openQL / openOL: Tensor Matrix Architecture for Unified Simulation of Physical Fields in 12D and 112D via GPU Tensor Cores
This document presents a theoretical and software-architectural proposal for a novel application programming interface (API) named openQL (designed for 12D macroscopic physics) and its fully expanded version, openOL (for the 112D quantum and molecular level). The architecture proposes abandoning traditional polygon geometry, as represented in OpenGL/Vulkan, and replacing it with a continuous multidimensional tensor field. By transforming quaternion and octonion algebra into unified block matrices, the complete computation of physical interactions (electromagnetism, thermodynamics, crystallography) is shifted to the hardware Tensor Cores of modern accelerators. The result is an asynchronous computational pipeline with zero-branching, opening the path to absolute real-time matter synthesis for industrial design, game engines, and materials engineering.
Authors
- Michal Mazgal
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-10-03
- DOI
- https://doi.org/10.5281/zenodo.23112560
- Primary Topic
- Model Reduction and Neural Networks
- Type
- preprint