Backward differentiation formulae with variable stepsize variable order for solving stiff delay differential equations / Nora Baizura Mohd Isa

Mohd Isa, Nora Baizura (2015) Backward differentiation formulae with variable stepsize variable order for solving stiff delay differential equations / Nora Baizura Mohd Isa. Masters thesis, Universiti Teknologi MARA.

Abstract

This thesis describes the development of predictor-corrector variable stepsize variable order based on backward differentiation formulae (BVSVO) method and direct predictor-corrector variable stepsize variable order based on backward differentiation formulae (DBVSVO) method for solving first order and special second order stiff delay differential equations respectively. The predictor and corrector formulae are represented in divided difference form. The developed methods are implemented using variable stepsize variable order technique. In varying the stepsize, the coefficients of the methods need to be recomputed at every step which will create extra computational cost. Thus, in order to reduce the computational cost, the coefficients of the methods are computed by a simple recurrence relation. In solving first order stiff delay differential equations using the BVSVO method, the numerical results are compared with non-stiff method, single-step method and multistep method.From the numerical results, the BVSVO method has shown the efficiency and reliability for solving first order stiff delay differential equations in terms of total number of steps, maximum error and average error. The DBVSVO method is used to solve special second order stiff delay differential equations directly without reducing to first order equations. We present some test examples to check an accuracy and efficiency of the DBVSVO method. For comparison purposes, the same set of test examples is reduced to a system of first order equations and solved using the BVSVO method. The numerical results for solving special second order stiff delay differential equations using the DBVSVO method is better compared with the BVSVO method in term of total number of function evaluations

Metadata

Item Type: Thesis (Masters)
Creators:
Creators
Email / ID Num.
Mohd Isa, Nora Baizura
UNSPECIFIED
Subjects: Q Science > QA Mathematics
Q Science > QA Mathematics > Mathematical statistics. Probabilities
Divisions: Universiti Teknologi MARA, Shah Alam > Faculty of Computer and Mathematical Sciences
Programme: Master of Science
Keywords: Backward Differentiation Formulae; Variable Stepsize Variable Order ;Stiff Delay Differential Equations
Date: 2015
URI: https://ir.uitm.edu.my/id/eprint/15139
Edit Item
Edit Item

Download

[thumbnail of TM_NORA BAIZURA MOHD ISA CS 15_5.pdf] Text
TM_NORA BAIZURA MOHD ISA CS 15_5.pdf

Download (1MB)

Fulltext

Fulltext is available at:
  • Library Terminal Workstation (Digital Format) - Accessible via UiTM Libraries
  • MyKM Portal

ID Number

15139

Indexing

|

Statistic

Statistic details