Two.

II Linear Independence

Linear Algebra, edition four

Jim Hefferon

http://joshua.smcvt.edu/linearalgebra
Definition and examples
Linear independence
1.4 Definition In any vector space, a set of vectors is linearly independent if
none of its elements is a linear combination of the others from the set.
Otherwise the set is linearly dependent .
Linear independence
1.4 Definition In any vector space, a set of vectors is linearly independent if
none of its elements is a linear combination of the others from the set.
Otherwise the set is linearly dependent .

Observe that, although this way of writing one vector as a combination

of the others
~s0 = c1~s1 + c2~s2 + · · · + cn~sn
visually sets off ~s0 , algebraically there is nothing special about that vector
in that equation. For any ~si with a coefficient ci that is non-0 we can
rewrite to isolate ~si .

~si = (1/ci )~s0 + · · · + (−ci−1 /ci )~si−1 + (−ci+1 /ci )~si+1 + · · · + (−cn /ci )~sn

When we don’t want to single out any vector we will instead say that
~s0 ,~s1 , . . . ,~sn are in a linear relationship and put all of the vectors on the
same side.
1.5 Lemma A subset S of a vector space is linearly independent if and only
if among its elements the only linear relationship c1~s1 + · · · + cn~sn = ~0 is
the trivial one, c1 = 0, . . . , cn = 0 (where ~si 6= ~sj when i 6= j) .
1.5 Lemma A subset S of a vector space is linearly independent if and only
if among its elements the only linear relationship c1~s1 + · · · + cn~sn = ~0 is
the trivial one, c1 = 0, . . . , cn = 0 (where ~si 6= ~sj when i 6= j) .
Proof If S is linearly independent then no vector ~si is a linear
combination of other vectors from S so there is no linear relationship where
some of the ~s ’s have nonzero coefficients.
If S is not linearly independent then some ~si is a linear combination
~si = c1~s1 + · · · + ci−1~si−1 + ci+1~si+1 + · · · + cn~sn of other vectors from S.
Subtracting ~si from both sides gives a relationship involving a nonzero
coefficient, the −1 in front of ~si . QED

So to decide if a list of vectors ~s0 , . . . ,~sn is linearly independent, set up

the equation ~0 = c0~s0 + · · · + cn~sn , and calculate whether it has any
solutions, other than the trivial one where all coefficents are zero.
Example This set of vectors in the plane R2 is linearly independent.
   
1 0
{ , }
0 1

The only solution to this equation

     
1 0 0
c1 + c2 =
0 1 0

is trivial c1 = 0, c2 = 0.
Example This set of vectors in the plane R2 is linearly independent.
   
1 0
{ , }
0 1

The only solution to this equation

     
1 0 0
c1 + c2 =
0 1 0

is trivial c1 = 0, c2 = 0.
Example In the vector space of cubic polynomials
P3 = { a0 + a1 x + a2 x2 + a3 x3 | ai ∈ R } the set { 1 − x, 1 + x } is linearly
independent. Setting up the equation c0 (1 − x) + c1 (1 + x) = 0 and
considering the constant term and linear term, leads to this system

c0 + c1 = 0
−c0 + c1 = 0

which has only the trivial solution.

Example The nonzero rows of this matrix form a linearly independent set.

2 0 1 −1
 
0 1 −3 1/2
 
0 0 0 5 
0 0 0 0

We showed in Lemma One.III.2.5 that in any echelon form matrix the

nonzero rows form a linearly independent set.
Example The nonzero rows of this matrix form a linearly independent set.

2 0 1 −1
 
0 1 −3 1/2
 
0 0 0 5 
0 0 0 0

We showed in Lemma One.III.2.5 that in any echelon form matrix the

nonzero rows form a linearly independent set.
Example This subset of R3 is linearly dependent.
     
1 −1 1
{ 1  ,  1  , 3  }
3 0 6

One way to see that is to spot that the third vector is twice the first plus
the second. Another way is to solve the linear system

c1 − c2 + c3 = 0
c1 + c2 + 3c3 = 0
3c1 + 6c3 = 0

and note that it has more than just the solution c1 = c2 = c3 = 0.

1.2 Lemma Where V is a vector space, S is a subset of that space, and ~v is
an element of that space, [S ∪ {~v }] = [S] if and only if ~v ∈ [S].
1.2 Lemma Where V is a vector space, S is a subset of that space, and ~v is
an element of that space, [S ∪ {~v }] = [S] if and only if ~v ∈ [S].

Example The book has the proof; here is an illustration. The span of this
set is the xy-plane.
   
2 0
P = { 0 , 2 } ⊂ R3
0 0
If we expand the set by adding a vector {~p1 , ~p2 , ~q } then there are two
possibilities.
           
2 0 3 2 0 0
P0 = { 0 , 2 , 2 } P1 = { 0 , 2 ,  0  }
0 0 0 0 0 −1

If the new vector is already in the starting span ~q ∈ [P] then the span is
unchanged [P0 ] = [P]. But if the new vector is outside the starting span
~q ∈/ [P] then the span grows [P1 ] ) [P].
1.3 Corollary For ~v ∈ S, omitting that vector does not shrink the span
[S] = [S − {~v }] if and only if it is dependent on other vectors in the
set ~v ∈ [S].
Example These two subsets of R3 have the same span
         
1 4 7 1 4
{ 2 , 5 , 8 } { 2 , 5 }
3 6 9 3 6

because in the first set ~v3 = 2~v2 − ~v1 .

 1.3 Corollary For ~v ∈ S, omitting that vector does not shrink the span
[S] = [S − {~v }] if and only if it is dependent on other vectors in the
set ~v ∈ [S].
Example These two subsets of R3 have the same span
         
1 4 7 1 4
{ 2 , 5 , 8 } { 2 , 5 }
3 6 9 3 6

because in the first set ~v3 = 2~v2 − ~v1 .

1.14 Corollary A set S is linearly independent if and only if for any ~v ∈ S, its
removal shrinks the span [S − { v }] ( [S].
Example This is a linearly independent subset of P3

S = { 1 + x, 1 − x, x2 }

Removal of any element, such as if we remove 1 − x to get Ŝ = { 1 + x, x2 },

will make the span smaller: [Ŝ] ( [S].
1.15 Lemma Suppose that S is linearly independent and that ~v ∈
/ S. Then the
set S ∪ {~v } is linearly independent if and only if ~v ∈
/ [S].
Example The book has the proof; here is an illustration. Consider this
linearly independent subset of P2 .

S = { 1 − x, 1 + x }

Its span [S] is the set of linear polynomials { a + bx | a, b ∈ R }. (To

check: consider a + bx = r1 (1 − x) + r2 (1 + x), which gives a linear
system with equations r1 + r2 = a and −r1 + r2 = b, having the solution
r2 = (1/2)a + (1/2)b and r1 = (1/2)a − (1/2)b.)
Here are two supersets.

S1 = S ∪ { 2 + 2x } S2 = S ∪ { 2 + x2 }
1.15 Lemma Suppose that S is linearly independent and that ~v ∈
/ S. Then the
set S ∪ {~v } is linearly independent if and only if ~v ∈
/ [S].
Example The book has the proof; here is an illustration. Consider this
linearly independent subset of P2 .

S = { 1 − x, 1 + x }

Its span [S] is the set of linear polynomials { a + bx | a, b ∈ R }. (To

check: consider a + bx = r1 (1 − x) + r2 (1 + x), which gives a linear
system with equations r1 + r2 = a and −r1 + r2 = b, having the solution
r2 = (1/2)a + (1/2)b and r1 = (1/2)a − (1/2)b.)
Here are two supersets.

S1 = S ∪ { 2 + 2x } S2 = S ∪ { 2 + x2 }

On the left, adding a linear polynomial just adds “repeat information” so

[S1 ] = [S] and S1 is linearly dependent.
The right, with “new information,” enlarges the span [S2 ] = P2 ) [S]
and the new set S2 is also linearly independent. (To check this, use
a + bx + cx2 = r1 (1 − x) + r2 (1 + x) + r3 (2 + x2 ) to get a linear system with
solution r3 = c, r2 = (1/2)a + (1/2)b and r1 = (1/2)a − (1/2)b − c.)
1.17 Corollary In a vector space, any finite set has a linearly independent
subset with the same span.

The book has a proof. Instead, consider the example on the next slide.
Example Consider this subset of R2 .
         
2 3 1 0 1
S = {~s1 ,~s2 ,~s3 ,~s4 ,~s5 } = { , , , , }
2 3 4 −1 −1

The linear relationship

           
2 3 1 0 1 0
r1 + r2 + r3 + r4 + r5 =
2 3 4 −1 −1 0

gives a system of equations.

2r1 + 3r2 + r3 + r5 = 0 −ρ1 +ρ2 2r1 + 3r2 + r3 + r5 = 0

−→
2r1 + 3r2 + 4r3 − r4 − r5 = 0 + 3r3 − r4 − 2r5 = 0
Parametrize by expressing the leading variables r1 and r3 in terms of the
free variables.
       
r1 −3/2 −1/6 −5/6
r   1   0   0 
 2      
{ r3  =  0  r2 +  1/3  r4 +  2/3  r5 | r2 , r4 , r5 ∈ R }
       
       
r4   0   1   0 
r5 0 0 1
Parametrize by expressing the leading variables r1 and r3 in terms of the
free variables.
       
r1 −3/2 −1/6 −5/6
r   1   0   0 
 2      
{ r3  =  0  r2 +  1/3  r4 +  2/3  r5 | r2 , r4 , r5 ∈ R }
       
       
r4   0   1   0 
r5 0 0 1

Set r5 = 1 and r2 = r4 = 0 to get r1 = −5/6 and r3 = 2/3,

           
5 2 3 2 1 0 1 0
− · +0· + · +0· +1· =
6 2 3 3 4 −1 −1 0

showing that ~s5 is in the span of the set {~s1 ,~s3 }.

Similarly, setting r4 = 1 and the other parameters to 0 shows ~s4 is in the
span of the set {~s1 ,~s3 }. Also, setting r2 = 1 and the other parameters to 0
shows ~s2 is in the span of the same set.
So without shrinking the span we can omit the vectors ~s2 , ~s4 , ~s5
associated with the free variables. The set of vectors associated with the
leading variables, {~s1 ,~s3 }, is linearly independent and so we cannot omit
any members without shrinking the span.
1.19 Corollary A subset S = {~s1 , . . . ,~sn } of a vector space is linearly
dependent if and only if some s~i is a linear combination of the vectors ~s1 ,
. . . , ~si−1 listed before it.
1.19 Corollary A subset S = {~s1 , . . . ,~sn } of a vector space is linearly
dependent if and only if some s~i is a linear combination of the vectors ~s1 ,
. . . , ~si−1 listed before it.
Proof Consider S0 = { }, S1 = { s~1 }, S2 = {~s1 ,~s2 }, etc. Some index i > 1 is
the first one with Si−1 ∪ {~si } linearly dependent, and there ~si ∈ [Si−1 ].
QED
Linear independence and subset
1.20 Lemma Any subset of a linearly independent set is also linearly
independent. Any superset of a linearly dependent set is also linearly
dependent.
Proof Both are clear. QED
Linear independence and subset
1.20 Lemma Any subset of a linearly independent set is also linearly
independent. Any superset of a linearly dependent set is also linearly
dependent.
Proof Both are clear. QED

This table summarizes the cases.

Ŝ ⊂ S Ŝ ⊃ S
S independent Ŝ must be independent Ŝ may be either
S dependent Ŝ may be either Ŝ must be dependent

An example of the lower left is that the set S of all vectors in the space
R2 is linearly dependent but the subset Ŝ consisting of only the unit vector
on the x-axis is independent. By interchanging Ŝ with S that’s also an
example of the upper right.