Abstract
In the realm of science and technology research, solving the problem of large matrix inversions is a matter of common interest We introduce a parallel approach to calculate large matrix inversions in this paper. A substantial amount of memory is needed to deal with large matrixes. Therefore, we considered an algorithm to optimize the memory distribution in a Cloud Computing system. We used the Gauss-Jordan algorithm for our research. This algorithm has a substantial regional memory access tendency; therefore, we focused on reducing it. To do this, we divided the matrixes according to the number of workers, and we divided the original matrix into one direction horizontally and one direction vertically. Using the Gauss-Jordan algorithm, the calculation step is increased according to the size of the matrix. Also, the previous steps of the calculation results are used in the following steps. In order to perform the above processes, we used a parallel scheduler, and this parallel scheduler managed the all workers; it also synchronized each worker's calculations. Being that our research is focused on solving large matrix inversions, we decided to experiment in a Cloud Computing system.
Original language | English |
---|---|
Pages (from-to) | 2833-2843 |
Number of pages | 11 |
Journal | Information |
Volume | 15 |
Issue number | 7 |
State | Published - Jul 2012 |
Keywords
- Algorithm
- Cloud computing
- Matrix inversion
- Parallel computing