CHE 374 LEC # 3 Number representation, Approximations and Errors

Numbers
• We distinguish between integer numbers
and real numbers
324 is an INTEGER number
CHE 374 Computational Methods 1.426
Π
is a REAL number
is a REAL number

in Engineering
• In computers, numbers are represented
Number representation, approximation, error either as INTEGERS (fixed point numbers)
and error definitions or as FLOATING POINT numbers )
1.426 (or 1.426e0 or 0.1426e+1 ….)

1 2

• Computers are digital machines. They

store numbers as a string of zero’s and Decimal (base 10) Representation
ones
102 101 100 10-1 10-2 10-3
Example INTEGERS represented in a 16 bit system
7 5 4 . 3 5 6
1 = 0000000000000001 =20 7x102 5x101 4x100 3x10-1 5x10-2 6x10-3
2 = 0000000000000010 =21
754.356
3 = 0000000000000011 =20+21
128 = 0000000010000000 =27
131 = 0000000010000011 =27+21+20
Representation of the number 754.356 in the decimal system

3 4

Computational Methods in Engineering 1

CHE 374 LEC # 3 Number representation, Approximations and Errors

Binary (base 2) Representation Storing a number in computer memory

22 21 20 2-1 2-2 2-3

1 0 0 . 1 0 1
1x22 0x21 0x20 1x2-1 0x2-2 1x2-3
1x4 0x2 0x2 1x0.5 0x0.25 1x0.125 Computer representation of a real number
4.625

Representation of the number 4.625 in the binary system 32 bit computer

5 6

Floating point representation Floating Point Representation

• Fractional quantities are typically represented in
• Floating point representation allows both
computer using “floating point” form, e.g.,
fractions and very large numbers to be
expressed on the computer. However,
Integer part – Floating point numbers take up more room.
exponent
[Link] – Take longer to process than integer numbers.
mantissa Base of the number system
used
– Round-off errors are introduced because
mantissa holds only a finite number of
significant figures.

7 8

Computational Methods in Engineering 2

CHE 374 LEC # 3 Number representation, Approximations and Errors

Chopping and Rounding Another example

Example:
e = 2.718281828459046 to be stored on a base-10
>> x = 29/1300
system carrying 7 significant digits. x = 0.02230769230769
e=2.718281 chopping error >> y = 29 - 1300 * x
et = 8.284590453655483e-007 y = 3.552713678800501e-015
The round-off error occurs when x is stored, 29/1300
If rounded
e=2.718282
et = -1.715409543301405e-007
• Some machines use chopping, because rounding adds
to the computational overhead. Since number of
significant figures is large enough, resulting chopping
error is negligible.
9 10

Errors in Computation Truncation errors

• Errors caused by use of truncated (shortened)
solution from an infinite series solution
• How confident we are in our
approximate result? We have already seen the approximation to the first derivative:
• The real question is “how much error
dv v v(tt 1 )  v (ti )
can we tolerate in our calculation” 
dt t

tt 1  ti

Can we estimate the error??

dv v (ti1 )  v(ti ) (1)

 e
dt ti1  ti
YES WE CAN !!!
11 USE Taylor series expansion of v(ti+1) around ti to find e 12

Computational Methods in Engineering 3

CHE 374 LEC # 3 Number representation, Approximations and Errors

1 1
v(t i  t )  v(t i 1 )  v(ti )  v t  v  t   v  t   ...
2 3

2! 3!

v(t i1 )  v (ti )  v t  R1

Dividing by ∆t and rearranging

v(t i1 )  v(ti ) dv(ti ) v(t i1 )  v(ti ) R
 v   R1 / t   1
or t dt t
  (t )
dv v (t i 1 )  v (ti ) R first order truncation
  1 approximation error
t
dt  (
  t )
1st order truncation
approximation error

Comparing with equation (1), we deduce that

1 d 2v 1 3
2 d v
e   t  2  t  3  ...  t 
2 dt 6 dt 13 14

Other sources/types of errors Error Definition

• Model errors – errors due to inadequate
representation of the physical model or parts True Value = Approximation + Error
of the model
et =true error= True value – Approximation
• Blunders and Mistakes - errors caused by
{may be positive or negative (+/-) }
malfunctions of the computer or human
imperfection.
Absolute error = |True value – Approximation|
• Data / input uncertainty - errors resulting from {always positive}
the accuracy and/or precision of the data.

15 16

Computational Methods in Engineering 4

CHE 374 LEC # 3 Number representation, Approximations and Errors

Error Definition Error definition

• In computations, we usually do not know the true
true error value. We can use the approximate value
relative error, er 
true value
 i n a n it e r a t iv e p r o c e s s :

true error ea
percent relative error, er %  100
true value
xi 1  xi
ea %   100%
xi 1
17 18

Q: What is the speedometer

Significant Figures
reading?
• Number of significant figures indicates
precision. Significant digits of a number are
those that can be used with confidence, e.g.,
the number of certain digits plus one
estimated digit.
35,700 How many significant figures?

3.57 x 104 3
3.570 x 104 4
3.5700 x 104 5
19 20

Computational Methods in Engineering 5

CHE 374 LEC # 3 Number representation, Approximations and Errors

Significant Figures Accuracy and Precision

Zeros are sometimes used to locate the

decimal point not significant figures.
0.00001753 4
0.0001753 4
0.001753 4

21 22

Accuracy and Precision

• Accuracy. How close a computed or measured
value is to the true value
• Precision (or reproducibility). How close a
computed or measured value is to previously
computed or measured values.
• Inaccuracy (or bias). A systematic deviation
from the actual value.
• Imprecision (or uncertainty). Magnitude of
scatter.
23

Computational Methods in Engineering 6