Type-Based Approaches to Rounding Error Analysis

Ariel Eileen Kellison · 2025

Read Type-Based Approaches to Rounding Error Analysis by Ariel Eileen Kellison (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-14598.txt.

Introduction

Ariel Eileen Kellison's 2025 dissertation builds programming languages that track rounding error inside their type systems. Floating-point arithmetic quietly distorts results, and checking the damage usually takes separate, slow tools.

The first language, NumFuzz, handles forward error: its types state quantitative error bounds, enforced by a linear discipline plus a grading mechanism that tallies accumulation. A prototype infers the bounds automatically, matching rival tools while often running much faster. The second language, Bean, treats backward error, asking how much the input would need to change to make the computed answer exact. Standard matrix algorithms written in Bean carry fine-grained bounds, again inferred automatically and matching published worst-case results.

Catalog record

CREATORAriel Eileen Kellison
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE402 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-14598

Browse the full library · About the Tower of Babel