Scribd
Upload
25 views

ADMATH Module 1

Numerical methods provide approximate solutions to mathematical problems that cannot be solved analytically. There are a few key characteristics of numerical solutions: [1] They are only available at discrete points, not continuously; [2] They involve trial and error processes that can result in unstable solutions if step sizes are improper; [3] They result in errors. Numerical methods convert continuous data to discrete data through approximation in order to solve problems using computers in an algebraic manner. The accuracy of numerical solutions depends on factors like the size of increments and can be measured by examining the true error, approximate error, and relative approximate error. Sources of error include truncation from dropping higher order terms and round-off errors from number limitations in computers.

Uploaded by

MLNDG boys
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
25 views

ADMATH Module 1

Numerical methods provide approximate solutions to mathematical problems that cannot be solved analytically. There are a few key characteristics of numerical solutions: [1] They are only available at discrete points, not continuously; [2] They involve trial and error processes that can result in unstable solutions if step sizes are improper; [3] They result in errors. Numerical methods convert continuous data to discrete data through approximation in order to solve problems using computers in an algebraic manner. The accuracy of numerical solutions depends on factors like the size of increments and can be measured by examining the true error, approximate error, and relative approximate error. Sources of error include truncation from dropping higher order terms and round-off errors from number limitations in computers.

Uploaded by

MLNDG boys
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

NUMERICAL METHODS

Introduction

Sometimes we can find nice, or simple, or exact solutions to equations.

Linear equations and quadratic equations, for example, can be solved exactly.

But other types of equations can be much more difficult to solve exactly. In fact,

sometimes it can be impossible to write down an exact expression for a solution.

When equations are difficult to solve, we can resort to approximate numerical

methods to find a solution. It is sometimes more efficient to find an approximate

answer. Numerical methods are techniques by which the mathematical problems

involved with the engineering analysis cannot readily or possibly be solved by

analytical methods.

There are a number of unique characteristics of numerical solution

methods in engineering analysis. Following are just a few obvious ones:

1. Numerical solutions are available only at selected (discrete) solution points,

but not at all points covered by the functions as in the case with analytical

solution methods.

2. Numerical methods are essentially “trail-and-error” processes. Typically,

users need to estimate an initial solution with selected increment of the

variable to which the intended solution will cover. Unstable numerical

solutions may result from improper selection of step sizes (the incremental

steps) with solutions either in the form of “wild oscillation” or becoming

unbounded in the trend of values.


3. Most numerical solution methods result in errors in the solutions.

Discrete Algebra

Like the human brain, computers are algebraic machines. They can

perform a finite number of elementary arithmetical and logical operations – and

that’s just about all! Conceptual understanding and numerical solution of

differential equations (an analytic construct!) can be accomplished only by

translating ‘analysis’ into ‘algebra’. And numerical data can be discrete or

continuous. Discrete data can only take certain values while continuous data

would (literally) take forever to count. In an algebra or calculus class, you might

have found a particular set of numbers (maybe the set of numbers in the range

of a function). You would represent this set as an interval: [0, ∞] is the range of

𝑓(𝑥) = 𝑥 2 since the set of outputs of the function are all real numbers 0 and

greater. This set of numbers is NOT discrete. The numbers in the set are not

separated by much at all. In fact, take any two numbers in the set and there are

infinitely many more between them which are also in the set. In the discrete

algebra, continuous data can be converted to a discrete data through

approximation.

Accuracy

Measurements and calculations can be characterized with regard to their

accuracy and precision. Accuracy refers to how closely a value agrees with the

true value. In many numerical methods, the calculations are executed in an

iterative manner until a desired accuracy is achieved. We can test this accuracy
of the approximation with a simple function where the derivative is easily

calculated. The accuracy of the solution is largely depending on the size of the

increments of the variable selected for the solutions.

Example:

Start at one value of x then change its value in small increment. A

change in the sign of f(x) indicates that there is a root within the last

increment.

Errors

Since numerical solutions are an approximation, and since the computer

program that executes the numerical method might have errors, a numerical

solution needs to be examined closely.

Measuring Errors

In any numerical analysis, errors will arise during the calculations. To be

able to deal with the issue of errors, we need to

1. identify where the error is coming from, followed by

2. quantifying the error, and lastly

3. minimize the error as per our needs.

There are four ways to in measuring errors in numerical solutions.

a. True Error

True error denoted by Et is the difference between the true value (also

called the exact value) and the approximate value.


𝑇𝑟𝑢𝑒 𝐸𝑟𝑟𝑜𝑟 (𝐸𝑡 ) = 𝑇𝑟𝑢𝑒 𝑣𝑎𝑙𝑢𝑒 − 𝐴𝑝𝑝𝑟𝑜𝑥𝑖𝑚𝑎𝑡𝑒 𝑣𝑎𝑙𝑢𝑒

b. Approximate Error

Approximate error is denoted by Ea and is defined as the difference

between the present approximation and previous approximation.

𝐴𝑝𝑝𝑟𝑜𝑥𝑖𝑚𝑎𝑡𝑒 𝐸𝑟𝑟𝑜𝑟(𝐸𝑎 ) = 𝑃𝑟𝑒𝑠𝑒𝑛𝑡 𝐴𝑝𝑝𝑟𝑜𝑥𝑖𝑚𝑎𝑡𝑖𝑜𝑛 − 𝑃𝑟𝑒𝑣𝑖𝑜𝑢𝑠 𝐴𝑝𝑝𝑟𝑜𝑥𝑖𝑚𝑎𝑡𝑖𝑜𝑛

c. Relative approximate error

Relative approximate error is denoted by (∈a ) and is defined as the ratio

between the approximate error and the present approximation.

𝐴𝑝𝑝𝑟𝑜𝑥𝑖𝑚𝑎𝑡𝑒 𝐸𝑟𝑟𝑜𝑟 (𝐸𝑎 )


𝑅𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑎𝑝𝑝𝑟𝑜𝑥𝑜𝑚𝑎𝑡𝑒 𝑒𝑟𝑟𝑜𝑟 (∈a ) =
𝑃𝑟𝑒𝑠𝑒𝑛𝑡 𝐴𝑝𝑝𝑟𝑜𝑥𝑖𝑚𝑎𝑡𝑖𝑜𝑛

Sources of Errors

There are two sources of errors that are inherent with numerical solutions:

a. Truncation errors – Because of the approximate nature of numerical

solutions, they often consist of lower order terms and higher order terms.

The latter terms are often dropped in the computations for the sake of

computational efficiency, resulting in error in the solution. For example,

consider the Taylor series expansion.

b. Round-off errors –Most digital computers handle either numbers with 7

decimal points, or 14 decimal points in numerical solutions. In the case of

32-bit computer with double precision (i.e. 14 decimal points length


numbers), any number after the 14th decimal point will be dropped. This

may not sound like a big deal, but if a huge number of operations are

involved in the computation, such error can accumulate and result in

significant error in the end results.

You might also like