Please use this identifier to cite or link to this item:
Parallel Processing in Numerical Integration
- Numerical Integration is a method of computing an approximation of the area under the curve of a function. The area under the curves is approximated the sum of the area of the subinterval rectangles. The difference between the sum of the rectangle’s area and the area under the curve is reduced increasing the number of subintervals. The trivial parallel implementation is that if p threads are available, the sum is sliced into p pieces. One interval is attached to each of intervals to compute a partial sum. The local sums are collected into one processor, which will know the answer. The computations are parallelized with OpenMP directives supported by Microsoft Visual Studio 2008. Serial time Ts is the time to run the program on one thread and parallel time T p is the time to run the program on p threads are obtained. By comparing Ts and Tp of different threads, the parallel processing is faster than serial processing.
Myint Myint Thein
- Universities Research Journal