Saved in:
| Main Author: | |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2026
|
| Online Access: | https://doi.org/10.5281/zenodo.19019977 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Table of Contents:
- <div>This work presents a lightweight rational approximation for the inverse square root function together with a Newton-compatible initialization strategy. The method is built around a low-order rational approximation on the normalized interval [1, 2], combined with exact power-of-two scaling to cover the practical range [0.25, 4].</div> <div> </div> <div>The construction is designed for low arithmetic cost, simple implementation, and predictable error behavior. On the normalized interval, the proposed approximation achieves an empirical maximum relative error of about 3.43e-4, while preserving a one-sided error structure that is useful for stable numerical handling. After a single Newton refinement step, the error is reduced to the 1e-7 scale.</div> <div> </div> <div>Rather than targeting formal optimality, this note focuses on practical usefulness: compact evaluation, explicit coefficients, and direct compatibility with fast inverse-square-root workflows. Potential applications include graphics, physics simulation, ray tracing, distance computation, and lightweight numerical kernels.</div>