Types of Errors. Numerical Methods.
Uploaded by
ScribdTranslationsTypes of Errors. Numerical Methods.
Uploaded by
ScribdTranslationsTypes of errors
Ms. Sarai
Numerical methods
Numerical errors are generated with the use of approximations for
represent the mathematical operations and quantities.
These include truncations that result from approximating
an exact mathematical procedure, and the rounding errors that result
to present approximately exact numbers.
ABSOLUTE ERROR
It is defined as the difference that exists between the exact value and its value.
calculated or rounded. The absolute error is not negative, because it
The definition was given in terms of absolute value. Thus, a sum
(collection) of errors always increases together, without decreasing.
Calculation of absolute error
To calculate the absolute error of a measurement, it is essential to know in
first place what value is considered as real. As a general rule that value is
the average of the values obtained when carrying out a number n of measurements in the
same conditions.
RELATED ERROR.
It is the quotient (the division) between the absolute error and the exact value. If we
Multiply by 100 to obtain the percentage (%) of error. Just like the
Absolute error can be positive or negative (depending on whether the absolute error is positive or negative)
because it can be due to excess or deficiency. it has no units.
Relative Error = |X – Xr| / |X|
ROUNDING ERROR
Rounding error. Almost all real numbers require, for their
decimal representation, of an infinity of digits. In practice, for your
management should only consider a finite number of digits in its
representation, proceeding to its determination through an appropriate
rounding.
Rounding errors are due to the fact that computers only store a
finite number of significant figures during a calculation. Computers
they perform this function in different ways. For example, if only they are saved
seven significant figures, the computer can store and use 'pi' as
“pi” = 3.141592, omitiendo los términos restantes y generando un error de
rounding.
TRUNCATION ERROR
Truncation errors are those that result from using a
approximation instead of an exact mathematical procedure. Also for
to obtain knowledge of the characteristics of these errors, it goes back to the
mathematical formulation widely used in numerical methods for
Express functions in polynomial form: Taylor series
SIGNIFICANT ERROR
Bear in mind that this type of error means that the number of digits
significant (that is, meaningful and valid) is sometimes
less than expected. It occurs more frequently when subtracting numbers
almost equal, but it can also happen when several numbers of magnitude and
different signs add up or when a relatively divisor is used
small.
PROPOGATED ERROR
It can be defined as the output error caused by an error in the input, assuming that
All intermediate calculations are carried out exactly (in particular, without rounding errors).
Include the evaluation of functions when the domain value is approximate, roots of
polynomials whose coefficients have been rounded or approximated, etc. Of course, in a
realistic situation all types of errors can intervene, so the output of a
the process will contain the propagated error plus the errors generated in the process.
xr denotes a rounded value of x;
ƒr(t) will indicate the rounded value of ƒ(t).
ƒ(tr) could be an infinite decimal number that must or would require rounding.
Formulas for Error Calculation
Absolute error | X – Xr |
Relative Error | X – Xr |
|X|
Modified Relative Error 2 | X – Xr |
|X| + |Xr|
Formulas for Error Calculation in Variable Functions
Absolute Propagated Error
Exact Absolute Error
| f(X) – f(Xr) | » | X – Xr | | f’(Xr) |
Relative Propagated Error | X – Xr | | f’(Xr) |
|f(X) |
Rounding Error | f(Xr) – fr(Xr)|
Total Evaluation Error f(X) – fr(Xr)
