Functional Expansion (FE) Algorithm Optimization - Benchmarking Data

Published: 27 March 2018| Version 2 | DOI: 10.17632/3rw29g8j9g.2
Contributor:
Brycen Wendt

Description

This data contains the effective computation time of different algorithms for evaluating terms in orthogonal polynomials series. Algorithms for the Legendre and Zernike polynomial series were evaluated. The underlying algorithms are described in an submitted journal article that is currently under review. The "answer" value of the data set has no external use. Instead, it was an accumulated value that was used throughout the benchmarking process to prevent over-aggressive compiler optimizations from removing the loops used for benchmarking. Conveniently, it provided a way to debug and validate the algorithm variations by comparison with each other. The source files were updated from v1 to correct errors in Zernike polynomials for $Z_7^{\pm3}$. These were propagated from: V. Lakshminarayanan and A. Fleck. Zernike polynomials: a guide [http://dx.doi.org/10.1080/09500340.2011.633763] The value '2' should be replaced by '21'.

Files

Steps to reproduce

This data was generated by compiling and running the included C source code. The program takes a single argument, the name of the output file. In this case "full.csv" was used. The program can also be configured to benchmark a subset of the included algorithms. This is done using the #define statements in lines 9-26 of main.c before compilation.