Parallel block methods for solving higher order ordinary differential equations directly
Numerous problems that are encountered in various branches of science and engineering involve ordinary differential equations (ODEs). Some of these problems require lengthy computation and immediate solutions. With the availability of parallel computers nowadays, the demands can be achieved. How...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English |
| Published: |
1999
|
| Subjects: | |
| Online Access: | http://psasir.upm.edu.my/id/eprint/8652/1/FSAS_1999_4_IR.pdf |
| _version_ | 1796967573104885760 |
|---|---|
| author | Omar, Zurni |
| author_facet | Omar, Zurni |
| author_sort | Omar, Zurni |
| collection | UPM |
| description | Numerous problems that are encountered in various branches of science and
engineering involve ordinary differential equations (ODEs). Some of these problems
require lengthy computation and immediate solutions. With the availability of
parallel computers nowadays, the demands can be achieved.
However, most of the existing methods for solving ODEs directly, particularly
of higher order, are sequential in nature. These methods approximate numerical
solution at one point at a time and therefore do not fully exploit the capability of
parallel computers. Hence, the development of parallel algorithms to suit these
machines becomes essential. In this thesis, new explicit and implicit parallel block methods for solving a
single equation of ODE directly using constant step size and back values are
developed. These methods, which calculate the numerical solution at more than one
point simultaneously, are parallel in nature. The programs of the methods employed
are run on a shared memory Sequent Symmetry S27 parallel computer. The
numerical results show that the new methods reduce the total number of steps and
execution time. The accuracy of the parallel block and 1-point methods is
comparable particularly when finer step sizes are used.
A new parallel algorithm for solving systems of ODEs using variable step size
and order is also developed. The strategies used to design this method are based on
both the Direct Integration (DI) and parallel block methods. The results demonstrate
the superiority of the new method in terms of the total number of steps and execution
times especially with finer tolerances.
In conclusion, the new methods developed can be used as viable alternatives
for solving higher order ODEs directly. |
| first_indexed | 2024-03-06T07:15:51Z |
| format | Thesis |
| id | upm.eprints-8652 |
| institution | Universiti Putra Malaysia |
| language | English English |
| last_indexed | 2024-03-06T07:15:51Z |
| publishDate | 1999 |
| record_format | dspace |
| spelling | upm.eprints-86522023-12-13T02:06:19Z http://psasir.upm.edu.my/id/eprint/8652/ Parallel block methods for solving higher order ordinary differential equations directly Omar, Zurni Numerous problems that are encountered in various branches of science and engineering involve ordinary differential equations (ODEs). Some of these problems require lengthy computation and immediate solutions. With the availability of parallel computers nowadays, the demands can be achieved. However, most of the existing methods for solving ODEs directly, particularly of higher order, are sequential in nature. These methods approximate numerical solution at one point at a time and therefore do not fully exploit the capability of parallel computers. Hence, the development of parallel algorithms to suit these machines becomes essential. In this thesis, new explicit and implicit parallel block methods for solving a single equation of ODE directly using constant step size and back values are developed. These methods, which calculate the numerical solution at more than one point simultaneously, are parallel in nature. The programs of the methods employed are run on a shared memory Sequent Symmetry S27 parallel computer. The numerical results show that the new methods reduce the total number of steps and execution time. The accuracy of the parallel block and 1-point methods is comparable particularly when finer step sizes are used. A new parallel algorithm for solving systems of ODEs using variable step size and order is also developed. The strategies used to design this method are based on both the Direct Integration (DI) and parallel block methods. The results demonstrate the superiority of the new method in terms of the total number of steps and execution times especially with finer tolerances. In conclusion, the new methods developed can be used as viable alternatives for solving higher order ODEs directly. 1999-12 Thesis NonPeerReviewed text en http://psasir.upm.edu.my/id/eprint/8652/1/FSAS_1999_4_IR.pdf Omar, Zurni (1999) Parallel block methods for solving higher order ordinary differential equations directly. Doctoral thesis, Universiti Putra Malaysia. Differential equations Parallel computers Parallel processing (Electronic computers) English |
| spellingShingle | Differential equations Parallel computers Parallel processing (Electronic computers) Omar, Zurni Parallel block methods for solving higher order ordinary differential equations directly |
| title | Parallel block methods for solving higher order ordinary differential equations directly |
| title_full | Parallel block methods for solving higher order ordinary differential equations directly |
| title_fullStr | Parallel block methods for solving higher order ordinary differential equations directly |
| title_full_unstemmed | Parallel block methods for solving higher order ordinary differential equations directly |
| title_short | Parallel block methods for solving higher order ordinary differential equations directly |
| title_sort | parallel block methods for solving higher order ordinary differential equations directly |
| topic | Differential equations Parallel computers Parallel processing (Electronic computers) |
| url | http://psasir.upm.edu.my/id/eprint/8652/1/FSAS_1999_4_IR.pdf |
| work_keys_str_mv | AT omarzurni parallelblockmethodsforsolvinghigherorderordinarydifferentialequationsdirectly |