| Home > Publications database > Algorithm 1062: Code Generation for Piecewise Chebyshev Approximation |
| Typ | Amount | VAT | Currency | Share | Status | Cost centre |
| APC | 1262.60 | 0.00 | EUR | 100.00 % | (Zahlung erfolgt) | ZB |
| Sum | 1262.60 | 0.00 | EUR | |||
| Total | 1262.60 |
| Journal Article | FZJ-2026-02134 |
;
2026
ACM
New York, NY
This record in other databases:
Please use a persistent id in citations: doi:10.1145/3805698 doi:10.34734/FZJ-2026-02134
Abstract: A Lipschitz continuous function f that has a high-precision, albeit slow reference implementation can beevaluated more rapidly by piecewise polynomials, obtained by Chebyshev approximation. The partition intosubdomains should be made such that the computation of subdomain index and reduced function argument isfast and introduces no rounding errors. A rigorous analysis of truncation and rounding errors shows thatthe maximum relative error can be chosen arbitrarily close to 2ε, except around zeroes of f . Alignment ofcoefficients minimizes cache loading times. Code generation and target algorithms are implemented in theopen source project ppapp, maintained at https://jugit.fz-juelich.de/mlz/ppapp, with a snapshot archived inthe Collected Algorithms of the ACM. It has been successfully applied to three real-valued functions in theopen source complex error function library libcerf.
Keyword(s): Information and Communication (1st) ; Instrument and Method Development (1st) ; Instrument and Method Development (2nd) ; Others (2nd)
|
The record appears in these collections: |