Chapter 1: Numerical Analysis and Introduction to Scientific Computing
1.1 Numerical Analysis: Its Objectives, Role, and Features
1.1.1 Mathematical Science and Numerical Analysis
Numerical analysis creates methods to find approximate solutions to mathematical
problems that cannot be solved exactly. It is used in applications like weather prediction
and simulating fluid flow in oil fields.
1.1.2 Computational Mathematics and Scientific Computing
Scientific computing combines numerical analysis, math, and programming to solve
practical problems, while computational mathematics focuses on teaching computers how
to solve mathematical equations using algorithms.
1.1.3 Computational Methods and Computers
Computational methods provide rules and formulas for solving numerical problems.
Computers perform millions of calculations using these methods, making them efficient
tools for solving complex problems.
1.1.4 Solving Numerical Problems
Solving numerical problems involves breaking the problem into smaller pieces, applying a
suitable method, and checking the results for accuracy. Example: Estimating the area of an
irregular shape using smaller rectangles.
1.2 Errors in Numerical Computation
1.2.1 Sources and Classification of Errors
Errors in numerical computation can be classified into three main types:
1. Round-off Errors: Occur when numbers are rounded during storage or computation (e.g.,
3.141592 is stored as 3.14).
2. Truncation Errors: Happen when a process is stopped early, such as summing only part of
an infinite series.
3. Data Errors: Result from inaccuracies in input data or measurements, such as recording
10.2 MPa instead of 10.25 MPa.
1.2.2 Errors and Significant Figures
Significant figures indicate the precision of a number. For example, 12.34 has 4 significant
figures, while 0.0056 has 2. Precision is essential to ensure accurate results without wasting
resources.
1.2.3 Error Estimation in Numerical Operations
Errors can be estimated using the following formulas:
1. Absolute Error (Ea): The difference between the true value (xt) and the approximate
value (xa):
Ea = |xt - xa|
2. Relative Error (Er): A comparison of the absolute error to the true value:
Er = Ea / |xt|
1.3 Error Analysis to Avoid Error Hazards
1.3.1 Understanding Error Propagation
When solving problems, small errors can grow larger as calculations continue. This is called
error propagation. Understanding how errors spread helps us design methods to minimize
their impact.
1.3.2 Conditioning of a Problem
Conditioning refers to how sensitive a problem is to small changes in its input. A well-
conditioned problem has outputs that do not change much even if the inputs vary slightly,
while an ill-conditioned problem amplifies small changes.
Example: Calculating the roots of a quadratic equation can be sensitive to errors if the
coefficients are nearly equal.
1.3.3 Stability of Numerical Methods
A stable numerical method produces results that are not overly affected by small errors
during calculations. Stability ensures reliable results even when errors occur at
intermediate steps.
1.3.4 Mitigating Error Hazards
To avoid errors from causing major issues:
1. Use well-conditioned methods when possible.
2. Ensure numerical methods are stable by testing them on known problems.
3. Use higher precision in calculations to reduce round-off and truncation errors.
