(c)2015 UM Math Dept

By-NC-SA 4.0 International License.

Math 217 Proof Techniques

Professor Karen Smith

This document lists some of the proof techniques we learn in Math 217.

1 General Techniques
G ENERAL TECHNIQUE TO GET STARTED : Restate clearly your hypotheses (givens) using sym-
bolic mathematical language.

Example: Show that if the kernel of a linear transformation T is trivial, then T is injective.
First Line: Assume ker T = 0.

T ECHNIQUE TO DISPROVE SOMETHING : Just give an explicit counterexample. It is fine (even

desirable) to take the simplest possible example.

Example: Prove or disprove that matrix multiplication is commutative.

Proof: False! State clearly what you
 will do.
  1
Let A = 0 1 and B = . State clearly your counterexample.
0
 
0 1
Then AB = [0] but BA = . Verify.
0 0

T ECHNIQUE TO PROVE SOMETHING EXISTS : Just give an explicit example. It is fine (even desir-
able) to take the simplest possible example.

Example: Show that there exists a non-zero matrix such that A2 = 0.

 
0 1
Proof: Let A = . Note A2 is the zero matrix.
0 0
P ROOF T ECHNIQUE : I NDUCTION . This often works for statements P (n) that are indexed by
whole numbers n. First prove the base case (usually case n = 1). Then show P (k) =⇒ P (k +1).

 N  
1 1 1 N
Example: Prove that = .
0 1 0 1
 on N
Proof: We induce 1 .  
1 1 1 1
Base Case: = .
0 1 0 1
 k  
1 1 1 k
Inductive Assumption: Assume = .
  0 1 0 1
1 1
Multiply both sides by to get
0 1
  k   
1 1 1 1 1 1 1 k
=
0 1 0 1 0 1 0 1
 k+1     
1 1 1 1 1 k 1 k+1
= . =
0 1 0 1 0 1 0 1
The proof is complete by induction.

T ECHNIQUE TO SHOW SOMETHING IS “ UNIQUE ”: assume there are two, do some math to show
they are the same.

Example: Show that the inverse of an invertible matrix is unique.

Proof: Let A be an invertible matrix. Name relevant objects.
Suppose that B and C are both inverses to A. Assume two.
So AB = BA = In and AC = CA = In . Write out what things mean.
So AB = AC (by substitution). Do some math.
Multiply both sides by B on the left to get B(AB) = B(AC). Do some math.
So (BA)B = (BA)C (associative property). Do some math.
So In B = In C and finally B = C.

P ROOF BY C ONTRAPOSITIVE : To “P implies Q,” it is logically equivalent to prove “Not Q implies

not P.” Sometimes this is easier.

T
Example: Let V → W be a linear transformation and let ~v1 , ~v2 , . . . , ~vd be vectors in V . Prove
that if T (~v1 ), . . . T (~vd ) are linearly independent, then so are ~v1 , ~v2 , . . . , ~vd .
Proof:
 We prove the contrapositive statement: Stating Contrapositive:
If ~v1 , ~v2 , . . . , ~vd are linearly dependent, then T (~v1 ), . . . T (~vd ) are linearly dependent.
We have a relation a1~v1 + a2~v2 + . . . ad~vd = 0 with at least one ai 6= 0. Write out meanings.
Apply T : T (a1~v1 + a2~v2 + . . . ad~vd ) = T (0) = 0. Do some math.
So a1 T (~v1 ) + a2 T (~v2 ) + . . . ad T (~vd ) = 0 because T is linear. Use definitions.
This is a non-trivial relation on T (~v1 ), . . . T (~vd ).
So T (~v1 ), . . . T (~vd ) are linearly dependent. QED.

2 General technique for inclusions, unions and intersections

T ECHNIQUE TO SHOW TWO SETS X AND Y ARE THE SAME : First show X ⊂ Y . Next show
Y ⊂ X.

Example: Show that (X ∪ Y )c = X c ∩ Y c .

S CAFFOLD :
We first show that (X ∪ Y )c ⊂ X c ∩ Y c .
MATH
We next show that X c ∩ Y c ⊂ (X ∪ Y )c .

T ECHNIQUE TO SHOW X ⊂ Y : Take an arbitrary element x of X and then do some math to see
that also x is in Y .

Example: Show that if T 2 = 0, then im T ⊂ ker T .

First Line: Take arbitrary ~x ∈ im T . The standard technique to show inclusion.
Next line: We need to show ~x ∈ ker T .

T ECHNIQUE TO SHOW W ⊂ X ∩ Y : Show separately that W ⊂ X and that W ⊂ Y .

Example: Prove that (X ∪ Y )c ⊂ X c ∩ Y c .

Proof: Take arbitrary x ∈ (X ∪ Y )c . The standard technique to show inclusion.
This means x ∈ / (X ∪ Y ). Write out what things mean.
So x ∈/ X, which means x ∈ X c . Showing each set separately.
/ Y , which means x ∈ Y c . Showing each set separately.
Also, x ∈
Thus x ∈ X c ∩ Y c . QED.

T ECHNIQUE TO SHOW A MATRIX IS WHAT YOU CLAIM IT IS : check each column is what you
claim by multiplication by ~ei using the Unusually Useful Lemma.
 T
Example: Let Rn → A
Rm be the map given by left multiplication by A. Show that if TA is the
zero map, then A is the zero matrix.
First line: Assume TA = 0. Restate givens.
Second line: This means TA (~x) = 0 for all ~x. Restate what things mean in precise, concise math.
In particular TA (~ei ) = A~ei = 0 so the i-th column is zero. The unusually useful lemma.
Since this holds for all i, all columns are zero, and so the matrix is zero.

3 Techniques to show injectivity, surjectivity, invertibility.

T ECHNIQUE TO SHOW A MAP f : X → Y IS INJECTIVE : Take two arbitrary x1 , x2 ∈ X. Assume
that f (x1 ) = f (x2 ). Then do some math to show that actually x1 = x2 .

T
Example: Show that if the kernel of a linear transformation V → W is trivial, then T is
injective.
Proof: Assume ker T = 0. Restate givens.
Suppose that T (~x) = T (~y ) for some ~x, ~y ∈ V . The standard technique to show injectivity.
Then 0 = T (~x) − T (~y ) = T (~x − ~y ) because T is linear. Do some math...
So ~x − ~y ∈ ker T . When stuck...how can I use my hypotheses?
Since ker T = 0, we deduce ~x − ~y = 0, so ~x = ~y . Thus T is injective.

Another technique is to make use of the following important and very useful Theorem: A
LINEAR TRANSFORMATION IS INJECTIVE IF AND ONLY IF ITS KERNEL IS ZERO . This leads to...

A NOTHER T ECHNIQUE TO SHOW A LINEAR TRANSFORMATION IS INJECTIVE : Just show the

kernel is trivial. Do this by taking an arbitrary element in the kernel, then showing it must be zero.

Example: Let T : R[x] → R[x] be the linear transformation sending f 7→ xf . Prove that T is
injective.
Proof:
It suffices to show ker T = 0. Standard technique: Just show the kernel is ZERO.
Let f = a0 + a1 x + · · · + an xn be an arbitrary element in ker T . Write out arbitrary element
This means T (f ) = xf = a0 x + a1 x2 + · · · + an xn+1 = 0. Write out what stuff means.
So a0 = a1 = · · · = an = 0.
So f = 0.
Since the kernel is trivial, the linear transformation T is injective.
T ECHNIQUE TO SHOW A MAP f : X → Y IS SURJECTIVE : Take arbitrary y ∈ Y , then do some
math to cook up some x ∈ X so that f (x) = y.

Example: Show that if A is an d × n matrix of rank d, then the linear transformation given by
left multiplication by A is surjective.
First line: Let A be a d × n matrix of rank d. Restate givens.
Next line: Take arbitrary ~y ∈ Rd . Technique for surjectivity: Take arbitary ~y in the target.
Next line: We need to find ~x ∈ Rn such that A~x = ~y . Restate what needs to be done.

A NOTHER TECHNIQUE TO SHOW SURJECTIVITY: Show that the image equals the target. For a
linear transformation, it is also enough to show the dimension of the image equal to the dimension
of the target (if this is finite dimensional).

 
1 2 0 4 5 0
Example: Prove the map given by left multiplication by the matrix 0 √π 1 4 12 0
0 17 0 4 5 1
is surjective.
Proof:
The image is spanned by the columns. Make use of Theorems.
Note that the columns 1, 3, 6 are ~e1 , ~e2 , ~e3 span all of R3 .
So the image is all of R3 .
This means the map is surjective.

S OME T ECHNIQUES TO SHOW A MAP f : X → Y IS BIJECTIVE ( OR INVERTIBLE ):

1. You can show BOTH f is injective AND f is surjective.

2. You can take arbitrary y ∈ Y and show that there is a unique x ∈ X such that f (x) = y.

3. You can define (or state what ) the inverse map is; be sure to verify that it is really the inverse
by checking the compositions in both orders.

Example for (3): Prove that rotation counterclockwise by θ is a bijective map of R2 .

Proof: The inverse is clearly clockwise through an angle of θ.
4 Techniques for Dealing with Kernel and Image
f
T ECHNIQUE TO WRITE ARBITRARY ELEMENT IN IMAGE : Given X → Y , then an arbitrary y in
the image of f can be written y = f (x) for some x ∈ X.

Example: Prove that im TA ⊂ W where TA is left multiplication by A.

First line: Take arbitrary ~y ∈ im TA . The standard technique to show inclusion.
Next Line: Write y = TA (~x) for some ~x. The standard way to write elements in image.
Next Line: So y = A~x. Rewrite what stuff means.

T ECHNIQUE TO SHOW ELEMENTS ARE IN KERNEL : To show ~x ∈ ker f , just show f (~x) = 0.

Example: Prove that if A2 = 0, then imTA ⊂ ker TA where TA is left multiplication by A.

First line: Assume A2 = 0. Restate givens concisely.
Then: Take arbitrary ~y ∈ imT . Standard technique to show X ⊂ Y .
Write ~y = TA (~x) for some ~x. Standard way to write elements of image.
This means ~y = A~x. Restate what things mean.
So A~y = A(A~x) = A2~x = 0~x = 0. Here we apply the transformation to check that we get 0.

5 Span, linearly independence and bases

T ECHNIQUE TO SHOW VECTORS SPAN : To show ~v1 , . . . , ~vd span W ,

1. Make sure each ~vi ∈ W (usually given)

2. Next show an arbitrary element of W is a linear combination of ~v1 , . . . , ~vd .

 
1
Example: Prove the plane defined by x + y + z = 0 is spanned by the vectors −1 and
  0
0
−1.
1
   
1 0
Note that both −1 and −1 satisfy x + y + z = 0. Step 1.
0 1 
x1
Take an arbitrary element  y1  on the plane. Starting Step 2.
z1
 This means  y1 = −(x
 that  1 + z
1 ). Rewrite
 what
 stuff
 means.
x1 x1 1 0
So  y1  = −x1 − z1  = x1 −1 + z1 −1 . Math.
z1 z1 0 1

T ECHNIQUE TO SHOW VECTORS ARE LINEARLY INDEPENDENT: To show ~v1 , . . . , ~vd are linearly
independent, write out a relation and show it must be trivial.

Example: Prove that the columns of an invertible matrix are linearly independent.
First Line: Let A be an invertible n × n matrix with columns C1 , . . . , Cn .1
Next: Suppose a1 C1 + · · · + an Cn =0 isa relation. Write out a relation.
a1
  a2 

This means that C1 C2 . . . Cn  ..  = ~0.
 
.
an
 
a1
 a2 
So A  ..  = 0.
 
.
an
 
a1
 a2 
Multiplying both sides by A−1 , we see  ..  = 0. Do some math to show the relation is trivial.
 
.
an
Thus our relation must be trivial. This means C1 , . . . , Cn are linearly independent.

T ECHNIQUE TO SHOW VECTORS ARE A BASIS : To show {~v1 , . . . , ~vd } are a basis for W , show
that they BOTH span W AND are linearly independent.

Example: Prove that ~e1 + ~e2 and ~e1 − ~e2 form a basis for R2 .
S CAFFOLD :
We first show ~e1 + ~e2 and ~e1 − ~e2 SPAN R2 .
We next show that ~e1 + ~e2 and ~e1 − ~e2 are LINEARLY INDEPENDENT.

A LTERNATIVE T ECHNIQUE TO SHOW VECTORS ARE A BASIS IF YOU KNOW THE DIMENSION :
Let {~v1 , . . . , ~vd } be a set of d vectors in a vector space W of dimension d. To show they form a
basis, you can prove either that they span W or that they are linear independent. You do not have
to check both if you know that you have the right number of vectors.2
2
The reason T HEOREM 3.3.4: A set of D vectors in a D-dimensional vector space V is linearly independent if
and only if it spans V .
 Example: Prove that ~e1 + ~e2 and ~e1 − ~e2 form a basis for R2 .
First Line: Since we know that the dimension of R2 is 2, it suffices to show that ~e1 + ~e2 and
~e1 − ~e2 SPAN R2 .
Alternate First Line: Since we know that the dimension of R2 is 2, it suffices to show that
~e1 + ~e2 and ~e1 − ~e2 are linearly independent.

6 Dealing with “and” and “or” Statements

T ECHNIQUE TO PROVE “AND ” STATEMENTS : To prove “P implies Q1 and Q2 ,” you need to
separately prove that “P implies Q1 ” and also that “P implies Q2 ”.
T IP : If you first show that “P implies Q1 ,” you are allowed to assume Q1 while showing that
“P implies Q2 .”

Example:. Prove that if TA : Rd → Rn is a linear transformation given by left multiplication

by a rank n matrix A, then TA is surjective and the kernal has dimension d − n.
Proof: We first show that TA is surjective. Showing P implies Q1 .
Recall that the dimension of the image of TA equals the rank of A. Making use of Theorems.
So the image is a subspace of Rn of dimension n.
This means im TA = Rn . Using characterization of surjective.
So TA is surjective.
We next show that ker TA has dimension d − n. Showing P implies Q2 .
Since dim Rd (source) is d and dim im TA = n, the Rank-Nullity theorem Use Theorems!
implies that dim ker TA = d − n. QED.

T ECHNIQUE TO PROVE “ OR ” STATEMENTS : To prove “P implies Q1 or Q2 ,” you can assume

“NOT Q1 and then show that “P implies Q2 ”. Alternatively, of course, you can assume “NOT Q2
and show that “P implies Q1 ”.
You should make sure you understand why this technique works. Note that if you need to show
“P implies Q1 OR Q2 ,” you would already be done if Q1 holds. That is the reason you might as
well assume Q1 doesn’t hold and just show “P and NOT Q1 together imply Q2 .”

Example:. Let ~v1 , . . . , ~vd be linearly independent elements in a vector space V . Then for
~ ∈ V , we have w
every w ~ ∈ Span {~v1 , . . . , ~vd } or {~v1 , . . . , ~vd , w}
~ is linearly independent.
Proof: Assume w~∈/ {~v1 , . . . , ~vd }. Technique for “or” statements.
We need to show that {~v1 , . . . , ~vd , w}
~ is linearly independent.
Let a1~v1 + · · · + ad~vd + ad+1 w
~ = 0 be a relation on {~v1 , . . . , ~vd , w}.
~
We need to show that the scalars a1 , . . . , ad+1 are all zero. Technique for linear independence.
 Case 1: ad+1 = 0. General Technique: handle different cases separately.
In this case, a1~v1 + · · · + ad~vd = 0 is a relation on {~v1 , . . . , ~vd },
and since {~v1 , . . . , ~vd } is independent, we have ai = 0 for i = 1, . . . d + 1.
Case 2: ad+1 6= 0. General Technique: handle different cases separately.
If ad+1 6= 0, we can re-arrange the relation to get General Technique: check case-by-case.
−a1 −ad
w
~= ~v1 + · · · + ~vd .
ad+1 ad+1

~ ∈ Span {~v1 , . . . , ~vd }.

This says that w
The proof is complete.

7 When all else fails

P ROOF T ECHNIQUE : C ONTRADICTION . Assume the conclusion is false and try to deduce an
obviously absurd statement.

Example: Let T : V → W be a linear transformation. Prove that if dim V > dim W , then T
is not injective.
Proof: Assume, on the contrary, that T is injective. Assume conclusion is false.
This means that ker T is trivial. Standard technique for handling injective linear transformation.
This means dim ker T = 0.
The Rank-Nullity Theorem says that dim V = dim imT + dim ker T , Make use of Theorems!
so in this case, dim V = dim imT .
Since imT ⊂ W , we also know that dim imT ≤ dim W .
Combining the last two sentences, we have dim V ≤ dim W .
This contradicts our hypothesis.
This contradiction completes the proof.

8 General Questions to Ask when you are stuck

1. What is the first line of the proof?

2. Can I scaffold the proof? For example, it is an “if and only if” statement, write out each
direction to be proved separately.

3. Have I named the relevant objects?

4. What am I trying to prove again?

5. Have I written out what things mean?

 6. How can I use the definition?

7. Have I used all the hypotheses?

8. What are some of the theorems related to the objects in the statement I am trying to prove?

9. Can I use rank-nullity? Theorem 3.3.4 is also very useful, can I use it?

10. What if restate the contrapositive? Can I get a start on the contrapositive statement?

11. Will induction work? Is there something I can induce on?

12. How would I start a proof by contradiction?