Introduction to Linear Transformations

Many branches of mathematics are concerned with studying functions with particular properties. For
example, single variable calculus is largely concerned with studying functions of one variable that are
differentiable. In linear algebra, the sort of function that we study is called a linear transformation, and
the goal of this handout is to explain what a linear transformation is.

If you’ve studied multivariable calculus, you’ve studied functions with different types of inputs and out-
puts. For example, you’ve studied functions like f ( x, y) = x2 + y2 , where the input is a pair of numbers
2 3
cos t
(x and y) and the output is one number. You’ve also studied parametric curves like ~r (t) = 4 sin t 5, where
t
the input (t) is a number and the output is a vector.

In linear algebra, we’ll study functions where the input and output are both vectors; such functions
are often also called transformations. (A linear transformation is a special kind of transformation, as we’ll
explain soon.) Our functions will often be named T (for “transformation”), and we will write T : Rm ! Rn
to mean that the inputs of T are vectors in Rm and the outputs of T are vectors in Rn . We call Rm the
domain of the function and Rn the codomain. For example, we could define a function T : R2 ! R3 by
2 3
✓ ◆ x1 + x2
x1
T = 4 x1 x2 5. This function has domain R2 (that is, the inputs are vectors in R2 ) and codomain
x2
x1 x2
R3 (the outputs are vectors in R3 ).

Visualizing a function T : Rm ! Rn

When you studied calculus, you probably visualized functions by drawing their graphs. This is too hard
to do for functions T : Rm ! Rn , so we use a different type of picture to visualize such functions.
✓ ◆ 
x1 x1
Example 1. Consider the function T : R2 ! R2 defined by T = .
x2 x2
✓ ◆  ✓ ◆ 
1 1 2 2
From the definition of this function, we see that, for example, T = and T = .
2 2 3 3
To visualize this, we draw a diagram showing both the domain and the codomain, with the domain on
the left. We can then draw some input vectors in the domain and the corresponding output vectors in the
codomain:

1
 
2
4  4
3 1
3 3
2
2 2
T
1 1

4 3 2 1 1 2 3 4 4 3 2 1 1 2 3 4
1 1
2 2 ✓ ◆
1
3 ✓ ◆ 3 T
2 2
4 T 4
3

For a better picture, we can draw more inputs and the corresponding outputs:

~v2
~v1 T (~v5 )
T (~v3 )
T
~v4

T (~v4 )

~v3
~v5 T (~v1 )
T (~v2 )

From this picture, you can probably see that T actually has a simple description in words: T (~v) is the
reflection of ~v over the x-axis in R2 . (Most functions T : Rm ! Rn do not have a simple geometric
interpretation like this, but the ones that do are valuable examples.) v
2 3
✓ ◆ 1
x1
Example 2. Consider the function T : R2 ! R3 defined by T = 4 x1 5. Again, we can visualize this
x2
x2
2 3 2 3
✓ ◆ 1 1
0
by drawing what T does to several vectors in the domain. For example, T = 405; of course, 405
1
1 1
is a vector in R3 , so we must visualize it in 3-dimensional space (which is hard to do on a 2-dimensional
sheet of paper!).

2
 ~v3 z

T (~v3 )
~v2 T
T (~v2 )

~v1
y
~v4
T (~v1 )
T (~v4 ) x

To see an interactive version of this picture (which will allow you to visualize the codomain more easily),
see the Handouts page of the course website. v

As you can imagine, this visualization is only useful in relatively simple cases. After all, it is impossible
to visualize R7 , so we have no hope of visualizing a function T : R7 ! R2 . Nonetheless, visualizing in
simple cases will help you build up your intuition so that you can understand what’s happening when
we study functions from R10 to R17 , even if you can’t visualize them directly!

Linear Transformations

Linear transformations are a special class of functions from Rm to Rn ; here is the definition.

Definition 3. A linear transformation is a function T : Rm ! Rn satisfying the following two properties:

1. T (~x + ~y) = T (~x ) + T (~y) for all ~x, ~y in Rm . (In words, we say that “T preserves addition”.)

2. T (k~x ) = kT (~x ) for all ~x in Rm and all scalars (real numbers) k. (In words, we say that “T preserves scalar
multiplication”.)

This definition probably seems quite abstract at the moment; an excellent habit when you read an abstract
definition is to think about some examples, so let’s do that now.
✓ ◆ 
x1 x + 3x2
Example 4. Consider the function T : R2 ! R2 defined by T = 1 . We’d like to decide
x2 2x1 x2
whether this is indeed a linear transformation. Reading Definition 3, we see that we need to decide
whether T preserves addition and scalar multiplication.

• Does T preserve addition? That is, is it true that T (~x + ~y) = T (~x ) + T (~y) for all ~x, ~y in R2 ?
 
x y
To decide this, let’s write everything out. If we write ~x = 1 and ~y = 1 , then
x2 y2
✓ ◆
x1 + y1
T (~x + ~y) = T
x2 + y2

( x1 + y1 ) + 3( x2 + y2 )
= by definition of T
2( x1 + y1 ) ( x2 + y2 )

3
 On the other hand,
✓ ◆ ✓ ◆
x1 y1
T (~x ) + T (~y) = T +T
x2 y2
 
x1 + 3x2 y + 3y2
= + 1
2x1 x2 2y1 y2

If we compare the two previous equations, we can see that T (~x + ~y) is indeed equal to T (~x ) + T (~y).
So, T preserves addition.

• Does T preserve scalar multiplication? That is, is it true that T (k~x ) = kT (~x ) for all ~x in R2 and all scalars
(real numbers) k?

x
Again, we simply write everything out. If ~x = 1 , then
x2
✓ ◆ 
kx1 kx1 + 3kx2
T (k~x ) = T =
kx2 2kx1 kx2

while

x1 + 3x2
kT (~x ) = k
2x1 x2

So, T (k~x ) = kT (~x ), and T preserves scalar multiplication.

Since T preserves both addition and scalar multiplication, T is a linear transformation .

Note: See the Mathematica page of the course website for applets that will help you visualize linear
transformations such as this one. v
2 3
✓ ◆ 1
x1
Example 5. Consider the function from Example 2, T : R2 ! R3 defined by T = 4 x1 5. Let’s
x2
x2
decide whether this is a linear transformation. Again, this means deciding whether T preserves addition
and scalar multiplication.

• Does T preserve addition? That is, is it true that T (~x + ~y) = T (~x ) + T (~y) for all ~x, ~y in R2 ?
 
x y
To decide this, let’s write everything out. If we write ~x = 1 and ~y = 1 , then
x2 y2
2 3
✓ ◆ 1
x1 + y1
T (~x + ~y) = T = 4 x1 + y1 5 by definition of T
x2 + y2
x2 + y2

4
 On the other hand,
✓ ◆ ✓ ◆
x1 y1
T (~x ) + T (~y) = T +T
x2 y2
2 3 2 3
1 1
= 4 x1 5 + 4 y1 5
x2 y2
2 3
2
= 4 x1 + y1 5
x2 + y2

If we compare the two previous equations, we can see that T (~x + ~y) is not equal to T (~x ) + T (~y). So,
T does not preserve addition.

Since T does not preserve addition, T is not a linear transformation .

Note: Here, we showed that T does not preserve addition in general; you could also show this by giving
2 3 2 3 2 3
✓ ◆ 1 ✓ ◆ 1 ✓ ◆ 1
1 0 1
a specific example. For example, T = 415, T = 405, and T = 415, so we see
0 1 1
0 1 1
✓  ◆ ✓ ◆ ✓ ◆
1 0 1 0
that T + 6= T +T . This single example is enough to ascertain that T does not
0 1 0 1
preserve addition, so T cannot be a linear transformation. v

In math, we often gain insight by looking at special cases of a general idea. We know that every linear
transformation preserves scalar multiplication; what does this say when the scalar is 0? If T : Rm ! Rn is
a linear transformation, then T (0~x ) = 0T (~x ) for any vector ~x in Rm . This can be written more simply as
T (~0) = ~0. That is, we have shown:

Fact 6. If T : Rm ! Rn is a linear transformation, then T (~0) = ~0.

This fact sometimes gives us a quick way to show that a given function is not a linear transformation. For
2 3
✓ ◆ 1
0
example, in Example 5, T = 05 6= ~0. Therefore, by Fact 6, we can immediately be sure that the
4
0
0
function T in Example 5 is not a linear transformation.

Be careful not to confuse Fact 6 with its converse(1) . It is not true that, if T : Rm ! Rn is a function
satisfying T (~0) = ~0, then T is a linear transformation.
✓ ◆ 
2 2 x1 x1 x2
Example 7. T : R ! R defined by T = satisfies T (~0) = ~0, but it is not a linear
x2 x1 + x2
transformation (can you explain why not?). v

(1) If you have a statement of the form, “If A, then B”, A is the hypothesis and B is the conclusion. The converse of a statement switches

the hypothesis and conclusion. For example, the converse of the statement “If a > 0, then a + 1 > 0” is “If a + 1 > 0, then a > 0.”