Approximations and Errors

These two are an integral part of human life. They are

everywhere and unavoidable.
Error = true value – approximate value
Understanding the concept of error is so important to the
effective use of numerical methods because every numerical
method involves an approximation.
For many applied engineering problems, we cannot obtain
analytical solutions. Therefore, we cannot compute exactly the
errors associated with our numerical methods. In these cases,
we must settle for approximations or estimates of the errors.
For example:
A speedometer from an automobile indicates that the car is traveling
between 48 and 49 km/h. One person might say 48.8, whereas
another might say 48.9 km/h. Therefore, because of the limits of this
instrument, only the first two digits can be used with confidence.
Abhishek Kumar Chandra 17 Saturday, November 13, 2021
 Estimates of the third digit (or
higher) must be viewed as
approximations.

Error came in variety of forms and sizes; some are avoidable

some are not.
For example:
Data conversion and round off error can not be avoided but
human error can be eliminated.
Although certain errors can not be eliminated completely, we
must at least know the bound of these error to make use of our
final solution.
It is therefore essential to know how errors arises, how they
grow during the numerical process and how they affect the
accuracy of a solution.
Abhishek Kumar Chandra 18 Saturday, November 13, 2021
 Total errors

Modeling error Inherent errors Numerical errors Blunders

Missing Human
information imperfection

Data Conversion Round off Truncation

errors errors errors errors

Measuring Numerical
Computing
method Method
Machine

Taxonomy of error
Abhishek Kumar Chandra 19 Saturday, November 13, 2021
 Data error: When data for a problem are obtained by some
experimental means and have a limited accuracy and precision.
This is due to some limitation of instruments and reading.
Conversion error: Arises due to limitation of the computer to
store the data exactly.
Round off error: When a fixed no. of digit are used to
represent exact no..
For example,
π = 3.141592653589793238462643…
Truncation error: Arises from using an approximation in place of
an exact mathematical procedure.

=
!
Inherent error: Error that are present in the data supplied to
the model.
Numerical error: This error introduced during the process of
implementation of numerical method.
Abhishek Kumar Chandra 20 Saturday, November 13, 2021
 Modeling Error
Mathematical models are the basis for numerical solution.
• Model formulated to represent physical process using
certain parameter involved in the situation.
• In many situation, it is impractical or impossible to include
all the real problem and therefore, certain simplifying
assumption are made.
For Example:
Developing a model for calculating the force acting on a falling
body.
 We may not be able to estimate the air resistance
coefficient (drag coefficient) properly or determine the
direction and magnitude of wind force acting on the body and
so on.
To simplify the model, we may assume that resistance is linearly
proportional to the velocity of the falling body or we may
assume that there is no wind force acting on the body.
Abhishek Kumar Chandra 21 Saturday, November 13, 2021
 Blunders
Error that are caused due to human imperfection. Such error
may cause a very serious disaster in the result.
It should be possible to avoid them to a large extent by acquiring
a sound knowledge of all aspect of the problem as well as
numerical process.
Human error can occur at any stage of the numerical processing
cycle . Some common types of errors are:-
• Lack of understanding of the problem
• Wrong assumption
• Overlooking of some basic assumption required for
formulating the model
• Selecting a wrong numerical method for solving the
mathematical model
• Wrong guessing of initial value
Abhishek Kumar Chandra 22 Saturday, November 13, 2021
 Numerical Errors (also known as procedural errors)
Introduced during the process of implementation of a
numerical method. They came in two form
Round off Truncation errors
Total numerical error is the summation of these two errors.
The total error can be reduced by devising suitable
techniques for implementing the solution
Round off Errors
When fixed no. of digit are used to represent exact no..
Since no. are stored at every stage of computation, round off
error is introduced at the end of every arithmetic operation.
Even though an individual round off error could be very small
the cumulative effect of a series of computation can be very
significant.

Abhishek Kumar Chandra 23 Saturday, November 13, 2021

 Rounding a number can be done in two ways
Chopping Symmetric rounding
Chopping:-
Extra digits are chopped. This is called truncating the number
Example:
42.7893 will be stored as 42.78 and the digit 93 will be
dropped.
Express the no. 42.7893 in floating point form as
x = 0.427893× 102
=(0.4278 + 0.000093) × 102
=(0.4278 + 0.93 × 10-4) × 102
This can be expressed in general form as
True x = (fx + gx × 10-d) × 10E fx is the mantissa
= fx × 10E + gx × 10E-d d is the length of mantissa
E is the exponent
= approximate x + error 0 ≤ g ≤1
x

Abhishek Kumar Chandra 24 Saturday, November 13, 2021

 Absolute error introduced depend on the following
• Size of the digit dropped
• Number of digits in mantissa
• Size of the number
Symmetric Rounding
Last retained significant digit is rounded up by 1 if the first
discarded digit is larger or equal to 5, otherwise the last
retained digit is unchanged.
Example :
42.7893 become 42.79
46.5432 become 46.54
Value of unrounded no. can be expressed as
True x = fx × 10E + gx × 10E-d
when gx ≤ 0.5 entire gx is truncated

Abhishek Kumar Chandra 25 Saturday, November 13, 2021

 Approximate x = fx × 10E
Error = gx × 10E-d gx < 0.5
When gx ≥ 0.5 last digit in the mantissa is increased by 1 and
therefore
Approximate x = fx × 10E + 10E-d
Error = (fx × 10E + gx × 10E-d) - fx × 10E + 10E-d
= (gx – 1) × 10E-d
So absolute error ≤ 0.5 × 10E-d
Symmetric rounding error is, at worst, one half the chopping
error.

Abhishek Kumar Chandra 26 Saturday, November 13, 2021

 Accuracy and Precision
The errors associated with both calculations and measurements
can be characterized with respect to their accuracy and
precision.

Accuracy refers to how closely

a computed or measured value
agrees with the true value.

Precision refers to how closely

individual computed or measured
values agree with each other.

Abhishek Kumar Chandra 27 Saturday, November 13, 2021

 Error
The numerical error is equal to the discrepancy between the
truth and the approximation.
True value = approximation + error
Absolute error Relative error
true error et = true value (xt) – approximation (xa)

Value of error may be + or – depends on value of xt and xa.

Most of the time magnitude of error is important not the sign.
So mode vale of error is known as absolute error.
et = │xt – xa│
In many cases this may not reflect its influence correctly as it
does not take account the order of magnitude of the value
under study.
For example: 1 gm is much more significant in the weight of 10
gm gold chain than in the weight of a bag of rice.
Abhishek Kumar Chandra 28 Saturday, November 13, 2021
 One way to account these magnitudes of the quantities is to
normalize the error to the true value, which is known as
relative or normalized error.
Relative or normalized error =

| − |
=
| |

The relative error can also be multiplied by 100 percent to

express it as
| |
= 100%
| |

Abhishek Kumar Chandra 29 Saturday, November 13, 2021

 It is often convenient to simplify the Taylor series by
defining a step size h = xi +1 – xi and expressing as

where the remainder term is now

The derivative of velocity of a falling parachutist is approximated by

a finite-divided-difference equation.
∆ ( )− ( )
≅ =
∆ −

Now let us truncate the series after the first derivative term:

Abhishek Kumar Chandra 30 Saturday, November 13, 2021

 Equation can be solved for

The truncation error associated with this approximation of the

derivative.

or

Thus, the estimate of the derivative has a truncation error of

order (ti+1 –ti ). In other words, the error of our derivative
approximation should be proportional to the step size.
Abhishek Kumar Chandra 31 Saturday, November 13, 2021
 Error Propagation
How errors in numbers can propagate through mathematical
functions.
For example: If we multiply two numbers that have errors, we
would like to estimate the error in the product.
• Functions of a Single Variable
Suppose that we have a function f(x) that is dependent on a
single independent variable x. Assume that is an
approximation of x. We, therefore, would like to assess the
effect of the discrepancy between x and on the value of the
function. We would like to estimate
∆ = −
The problem with evaluating ∆ is that is unknown
because x is unknown. We can overcome this difficulty if is
close to x and f ( ) is continuous and differentiable. If these
conditions hold, a Taylor series can be employed to compute
f(x) near f ( ), as in
Abhishek Kumar Chandra 32 Saturday, November 13, 2021
 Dropping the second- and higher-order terms and rearranging
yields

Or

where ∆ =| − | represents an estimate of the

error of the function and ∆ =| − | represents an
estimate of the error of x.
• Functions of More than One Variable

Abhishek Kumar Chandra 33 Saturday, November 13, 2021