An adaptive numerical integration code for a chain of transputers
The finite integration interval is splitted up into subintervals of equal length. On every subinterval a transputer calculates an approximation to this integral. This approximation uses an adaptive method based on the ten point Gaussian rule and the twenty-one point Kronrod extension [See also the QUADPACK routine DQAGS described by \textit{R. Piessens}, \textit{E. de Doncker-Kapenga}, \textit{C. W. Überhuber} and \textit{D. K. Kahaner}; QUADPACK, A subroutine package for automatic integration (1983; Zbl 0508.65005)]. Numerical tests (on examples of the book listed above) show on a chain of 32 T800s that if a large number of function evaluations is needed then linear speed-ups are achievable with any number of transputers.
This page was built for publication: An adaptive numerical integration code for a chain of transputers
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q750084)