WIDENER UNIVERSITY School of Engineering

Third Edition
Tradition & Excellence Since 1821 Mechanical Engineering

TOPIC

ENGR 680

02 Advanced Computational
Methods

Numbers and Vectors

© Tulong Zhu, All rights reserved.

Properties of Real Numbers, R
Real Number • Integers, rational numbers, and irrational numbers.
 The real numbers can be represented as points on the real
number line.
-1/2 2 
-4 -3 -2 -1 0 1 2 3 4

Associative ( x  y)  z  x  ( y  z)
( x  y)  z  x  ( y  z)
Commutative x y  yx
x y  y x

Identity x0  x
elements
1 x  x
Inverse x  ( x)  0
Element
if x  0, x  (1 / x)  1
02 - 2
Numbers in Computers: Binary Numbers
Binary • In computers, numbers are stored in the binary number system.
Number Only digits 0 & 1 are stored in memory to represent all possible
numbers.

Example 1= 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1

2= 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0

3= 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1

13= 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 1

02 - 3
Covert from Binary to Decimal: Integers
Integers • Integers: For a binary number dm-1 dm-2 … d2 d1d0, the decimal
value, N, is
m 1
N   di 2i  d 0  d1  2  d 2  2 2    d m 1  2 m 1
i 0

dm-1 dm-2 … d2 d1 d0
N=     
+ + +
2m-1 2m-2 … 22 21 20
 Since dm-1 , dm-2 , …, d2, d1 and d0 are either 1 or 0, N is the
sum of those 2i with a non-zero di , i = 1,  2,…,  m.

1 1 … 1 0 1
N= 2m-1+ 2m-2 …
+ 22 + 20
Example • (1101)2 = 1 + 0(21) + 1(22) + 1(23) = 13

0 0 0 0 1 1 0 1

02 - 4
23 + 22 + 20 = 8 + 4 + 1 = 13
Covert from Binary to Decimal: Fractions
Integers • Fractions: For a binary number .d-1 d-2 … d-(m+1) d-m, the decimal
value, N, is
m
N   d i 2 i  d 1  2 1  d  2  2  2    d  m  2  m
i 1

d-1 d-2 … d-(m+1) d-m

N=    
+ + +
2-1 2-2 … 2-(m+1) 2-m
 Since d-1 , d-2 , …, d-(m+1), and d-m are either 1 or 0, N is the
sum of those 2i with a non-zero di, i = 1,  2,…,  m.

1 1 … 1 0 1
N= 2-1 + 2-2 … + 2-(m+2)+ 20
Example • (.1101)2 = 1(2-1) + 1(2-2) + 0(2-3)+ 1(2-4) = 0.8125

1 1 0 1 0 0 0 0

02 - 5
2-1 + 2-2 + 2-4 =0.5+0.25+0.0625= 0.8125
Fixed Point Notation
Fixed Point • The binary point is not stored in the computer, but is remembered
Notation by the programmer (program), e.g., for a 8-bit number, the
program can assume the binary point is after the 5th bit.
Example The binary point is not stored

1 0 0 1 1 1 0 1
23 20 2-1 2-2 2-4 = 9.8125
Number • The biggest positive number is stored in the computer, when
Range every bit is stored with a digit 1.

1 1 1 1 1 1 1 1
27 26 25 24 23 22 21 2 0 = 255 ~28
 For a m-bit computer, the range of positive fixed-point
integers is about 2m

02 - 6
Fixed Point Notation: Limitations
Limit • If N bits are used, then only 2N things can be represented. 8 bits can
Precision represent 256 positive integers, 256 fractional numbers, and so on.

• For the 8-bit computer, with the binary point after the 4th bit, here is
a number line showing the 256 values that can be represented with
this fixed notation:

• The smallest value is zero (00000000); the largest value is 15.9375

(11111111). The smallest non-zero value is 0.0625. Every
represented value is a multiple of 0.0625.
Que  How many real numbers lie between 0 and 15.9375?

Conclu • For m bit, fractions include errors from 0 to 2-m.

Number • Very large numbers and very small fractions cannot be represented
Range with this limited bits (2m for a m-digit computer).
02 - 7
Floating Point Notation: Decimal
Decimal • In the decimal system we can get round the range problem using
Floating scientific notation or float point notation.
Notation
Scientific Floating Point
Notation Notation
1,245,000,000,000 = 1.2451012 = 0.12451013
0.00001245 = 1.24510-5 = 0.124510-4
-0.00001245 = -1.24510-5 = -0.124510-4

(e.g., 1.245 = + 0.1245  10-1 )

N.n =  M  BE

Decimal Value Sign bit Mantissa Base Exponent

02 - 8
Floating Point Notation: Binary Numbers
Binary • If we change the base into 2 in the previous slide, a decimal
Number number can always be written as:
N.n =  M  2 E
Decimal Value Sign bit Mantissa Base Exponent
 Only the Sign bit, Mantissa, the exponent and its sign need to
be stored in the computer.
Example • 5.75 = 5.75  20 = 101.11  20 = 101.11  20
= 2.875  21 = 10.111  21 = 10.111  21
= 1.4375  22 = 1.0111  22 = 1.0111  210 Binary
=0.71875  23 = .10111  23 = .10111  211 numbers
In a  Suppose we use 16-bits words, with 5 bits for the exponent:
Computer + . 10111  2 + 11

0 1 0 1 1 1 0 0 0 0 0 0 0 0 1 1
Sign bit Mantissa, M Sign of E Exponent, E
 The above scheme is for illustration only, the real storage
02 - 9 scheme in a computer is different (see IEEE 754).
Floating Point Notation: MATLAB©
Donald E. Floating-point computation is by nature inexact, and it is not
Knuth: difficult to misuse it so that the computed answers consist
almost entirely of 'noise'.
Range • Only a finite number of bits (32 for most PCs) are available for
a computer. The range of decimal numbers stored is also finite.
MATLAB© uses a double precision format (64 bits) to store
floating point numbers (with 16 significant decimal digits) and
32 bits to store integers, The ranges floating numbers is
 Floating numbers: ~ 1.7*10–308 to 1.7*10308 (or ~ 2–1024 to
21024) .
Precision • As with the binary fixed point notation, use a finite number of
bits in the binary floating point notation can also represent only
a finite number of decimal numbers in a computer.
• Therefore, a binary floating point representation in a computer
is always an approximation for most numbers.
• Floating point relative accuracy in MATLAB© is
>> eps=2.2204e-016 (machine dependent = the distance
from 1.0 to the next largest floating point number which is
smaller than 1.0).
02 - 10
Number in Computers: Errors

• Binary Floating Point computations will always bring errors. These

may include:
 Rounding
o Non-representable numbers: for example, the literal 0.1 cannot
be represented exactly by a binary floating-point number using
a finite number of bits.
o Rounding of arithmetic operations: for example 2/3 might
yield 0.6666667
 Absorption: 1×1015 + 1 = 1×1015
 Cancellation: subtraction between nearly equivalent operands
 Overflow / Underflow (number is too small or too big).
• Some axioms for real numbers may not exactly apply, e.g.,
(a + b) + c = a + ( b + c) ??
• Consequently, binary floating-point should not be used for financial
calculations, or indeed for any calculations where the results achieved
are required to match those which might be calculated by hand.

See Comp_error.m for errors

02 - 11
Useful MATLAB© Commands for Real Numbers

Command Format Example: -4.503, 3.491

abs Absolute value 4.5030, 3.4910
ceil Round toward infinity -4, 4
fix Round toward zero -4, 3
floor Round toward minus infinity -5, 3
round Round towards nearest integer -5, 3
sign Signum function -1, 1

See Real_num_command.m

02 - 12
Complex Numbers: Basic Concepts
Complex z  x  iy, i   1 Im
Num x = Re z = real part, z
y = Im z = imaginary part y
r
Magnitude r  z  x 2  y 2 (Length or Modulus) 
Argument,  x  (Multi-valued:  +2k)
  arctan 
 y  (Principal: [– , ])
Polar angle x Re

Complex z  x  iy
Conjugate  ( z )   ( z ), r ( z )  r ( z ) z
Basic z1  x1  iy1 , z2  x2  iy2
Operations
z1  z2  ( x1  x2 )  i ( y1  y2 )
2
z1z2  ( x1x2  y1 y2 )  i ( x1 y2  y1x2 ) z  zz
z1 z1 z2 z1z2 x1x2  y1 y2 y x x y
   i 1 2 1 2
z2 z2 z2 z2 2 x12  y12 x12  y12

z n  ??? e z  ??? z1 / n  ??? 

02 - 13
Complex Numbers: Polar Form
i Im
Preliminary e  cos  i sin 
z
De Moivre (cos  i sin  ) n  cos n  i sin n y
r
Polar Form z  x  iy  r (cos  i sin  )  re i 

 i
x Re
z  x  iy  r (cos  i sin  )  re

Remember? z n  ??? e z  ??? z1 / n  ??? 

z

 z n  (re i ) n  r n ein  r n (cos n  i sin n )

e z  e x iy  e x eiy  e x (cos y  i sin y )

z1 / n  (rei )1 / n  [rei (  2k ) ]1 / n

  2k   2k
 r1 / n (cos  i sin ), k  1,2, ,n-1
n n
02 - 14
Complex Numbers: MATLAB© Commands

Command Format
z=x+iy, z=x+jy Complex number
z=r*exp(i*theta) Polar form

abs(z) Magnitude z  x2  y 2
angle(z) Angle in radians (-, );  = tan–1(y/x)
conj(z) Complex conjugate x - iy
imag(z) Complex imaginary part y
real(z) Complex real part x
Plotting
compass(x,y) Draw complex numbers as arrows on polar plot
feather(x,y) Draw complex numbers as arrows on linear plot

See Complex_num.m

02 - 15
Vectors in High Dimensions
Definition • An n dimensional vector x  Rn (Rn is the set of of all vectors
with n real components) of real numbers is defined as
x = [x1, x2, …, xn] = x1e1 + x2e2 + … + xnen
e1 , e2 , … , en is a set of base (orthonormal) vectors

Addition x  y  x1  y1 x2  y2  xn  yn 

Scalar Mult kx  kx1 kx2  kxn 

Norm x  x12  x22    xn2 (also called the length)
x
Unit Vector u
x
n
Dot Product x  y  x, y  x1 y1  x2 y2    xn yn   xi yi
Inner Prod. i 1
2
x  xx
Cross Prod. x  y  ??? (Only applied to a 3D space)
02 - 16
Vectors: Orthogonal
Orthogonal • The nonzero vectors x and y are said to be orthogonal if the dot
product or inner product is zero, i.e., x·y = <x, y> = 0.
Orthonormal • Mutually orthogonal vectors in Rn with unit length are called
orthonormal.

 If m vectors v1 , v2 , … , vm in Rn are mutually

orthonormal, then
1 i  j
vi  v j  vi , v j   ij  
0 i  j
ij is the Kronecker delta.

 The three base vector i, j and k in a three dimensional space

are orthonormal.

 The n base vectors e1 , e2 , … , en in an n-dimensional space

are orthonormal.

02 - 17
Vectors: Linear Dependence
Linear • Let x1, x2, …, xm be a set of n-vectors in Rn. Then a linear
Combination combination of these vectors is a sum of the form
k1x1  k2x2    kmxm
where k1, k2, …, km are nonzero scalars.
Linear • Let x1, x2, …, xm be a set of vectors in Rn. Then the set is said
Dependence to be linearly independent if and only if none of the vectors can
be expressed as a linear combination of the remaining vectors.
Theorem • Let x1, x2, …, xm be a set of vectors in Rn . The set is linearly
independent if the vector equation
k1x1  k2x2    kmxm  0
is only true when k1= k2 = … = km = 0
 Linearly DEPENDENT: at least one non-zero scalar, ki,
exists such that
k1x1  k2x2    kmxm  0
Notes • The highest number of linear independent vectors in Rn is n.
• A mutually orthogonal set of vectors is linearly independent.
02 - 18
Vectors: MATLAB© Commands
Operation MatLab Command Math Form
Standard Addition (subs) xy [x1y1, x2y2, …, xnyn]
Linear Scalar Multpl k*x [kx1 , kx2 , … , kxn]
Algebra
Operation Transpose x’ Column Vector
Dot/inner Prod x*y’, dot(x,y) x1 y1 + x2 y2 + … + xn yn
Cross Prod (3D) cross(x,y) xy
Norm norm(x) x
Special Multiplication x.*y [x1 y1 , x2 y2 , …, xn yn]
element-by- Division x./y [x1/y1 , x2/y2 , …, xn/yn]
element
Exponentiation x.^m (m: scalar) [x1m, x2m, …, xnm]
operation in
MATLAB© Function f(x) [f (x1) , f (x2) , …, f (xn) ]
Augment [x y] [x1, x2, …, xn, y1, y2, …, yn]

See Vectors.m for vector operations.

02 - 19