Linear Algebra Harvard Notes Lecture 3

Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

Math 22a Lecture 3

1 Logistics
Office Hours : Monday 4-6 +7-9 in Sever 306
Prof Taubes: Tues 3-4 Sc. 504 Wed 4-5 Sc.504

2 Row reduction and row echelon forms


0 10 1
a11 a12 : : :: a1n b1
B a CB C
B 21 a22 : : : a2n CB b2 C
B    CB  C
@    : A@  A
am1 am2 : : : amn bm

Must do elementary row operations to the augmented matrix.


Aim is to get a matrix in Reduced Row Echelon Form (RREF)

2.1 Reduced Row Echelon Form


A  rref(A)
rref(A) :
0 1
1 0 5 0 b10
B C
B 0 1 4 0 b2 0 C
B C
B 0 :::
@ 1  C
A
0
0 : : :: 0 bn

 Left most nonzero entry in each row is 1.


 If the column is a pivot column has a leading 1 then all other entries are zero
 Leading 1 in previous row is to the left
 Rows that are all zero are at the bottom

Test for inconsistency :


System is inconsistent if any non-zero b's are in rows with all zeros.
i.e. the rightmost column in the augmented column is a pivot column

! 0 1
    1 5 1 1
5 ¡3 2 4 5 ¡3 2 4 5 0 ¡4 1 0 ¡ 20 2
10 ¡2 1 6
 0 4 ¡3 ¡2
 2 @ 3 1
A
0 4 ¡3 ¡2 0 1 ¡4 ¡2

1 1
x1 ¡ x3 =
20 2
3 1
x2 ¡ x3 = ¡
4 2

1 1
x1 = + x3 (1)
2 20
1 3
x2 = ¡ + x3 (2)
2 4

1
Algorithm:
 Pick a leftmost non-zero entry of A - that column will be the first pivot column.
 Repeat : find next leftmost non-zero entry of A - this will find all pivot columns that will be the
same throughout the different matrix forms
 Rearrange the rows such that it has a step-like structure
 Then divide each row with a non-zero entry by the leading entry.

2.2 Theorem: rref is bijective


Given A, this algorithm produces rref(A) and there is a unique rref(A).

Why is it unique?
Each elementary row operation is trivially reversible
Consider two rref matrices from the same matrix A:
The pivot columns remain the same - what about non-pivot columns?

3 Vectors
Column vector - a matrix with 1 column
Row vector - a matrix with 1 row

c.u
VECTOR ADDITION

u+v

v
u

0 1
x1
B x2 C
B C n
B  C2 R
@  A
xn

2
3.1 Properties of vectors
 Any two vectors can be added
0 1 0 1 0 1
x1 x10 x + x0
B  C B C B 1  1 C
@  A + B
@  C =
A @
B  C
A
xn 0 0
xn xn + xn

 c is any number
0 1 0 1
x1 c x1
@  A = @  A
C B C
cB
xn c xn

All vectors that are obtained from fv


~ 1; ~v2; : : : ; ~vpg are called linear combinations of v's.
The set of all such linear combinations is called span of fv
~ 1; : : : ; ~vpg
Examples of spans:
 Span of ~v - line through ~v
 Span of ~v1 and ~v2 - assuming not parallel, they get any point in the plane

You might also like