Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd
Upload
5 views

An Example Illustrating The Concepts of Basic Linear Algebra

Uploaded by

neeru572010
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
5 views

An Example Illustrating The Concepts of Basic Linear Algebra

Uploaded by

neeru572010
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

An Example illustrating the Concepts of Basic Linear Algebra

S. Kumaresan
School of Math. & Stat.
University of Hyderabad
Hyderabad 500 046
kumaresa@gmail.com

September 7, 2020

Abstract

There are results which are existential and theoretical in nature in a preliminary
course on linear algebra. (We have in mind the results such as the existence of a basis,
the extension of a linearly independent set in a vector space V to a basis of V , the proof
of rank-nullity theorem etc.) While it trains the young minds to think in abstract, it is
equally important to show the young minds how to turn those abstract arguments into
something concrete in practical situations. Also, often beginners think of linear maps
between Rn and Rm only as functions defined in terms of “formulas”. This is in spite of
the fact that they have seen a results which emphasize that a linear map T : V → W is
defined once we know the images of vectors in a basis of V under T . The purpose of this
article is to work out an example explicitly to show how the theory can be implemented
in concrete cases. It is also hoped that this example will show how to create linear maps
rather than checking a given map is linear!

Can we find a linear map T : R6 → R4 such that

ker T = U := {(x1 , . . . , x6 ) ∈ R6 : x1 = x2 = x3 and x4 + x5 + x6 } = 0,


Im T = V := {(y1 , . . . , y4 ) ∈ R4 : y1 + y2 + y3 + y4 = 0}?

The first thing we need to check is whether the Rank-Nullity theorem vetoes against the
existence of such a map. As we have learnt earlier, dim ker T = 3 while dim Im T = 3. Hence
dim R6 = 6 = dim ker T + dim Im T . So, the rank-nullity theorem does not rule out the
existence of such a linear map. Note that if we required ker T = {x ∈ R6 : x1 = x2 = x3 or if
we demanded that Im T = {y ∈ R4 : y1 = y2 , y3 = y4 }, then no such maps can exist. (Justify
this claim.)
Let us denote the standard basis of R6 by e1 , . . . , e6 and that of R4 by f1 , . . . , f4 . We
fix a basis of U : u1 := e1 + e2 + e3 , u2 = e4 − e6 and u3 := e5 − e6 . We fix vj := fj − f4 ,
j = 1, 2, 3 as a basis of V . The proof of Rank-Nullity theorem shows us how to construct a
linear map as required. We need to (i) extend the basis of U (by adding three vectors, say,
u4 , u5 , u6 ) to a basis of R4 , (ii) map {u1 , u2 , u3 } to 0 and (iii) map {u4 , u5 , u6 } bijectively
onto {v1 , v2 , v3 }. Finally write down the map explicitly using the standard coordinates and
verify it is as required.

1
Note that e1 + e2 + e3 lies in the span of {e1 , e2 , e3 }. So to find a basis that includes
e1 + e2 + e3 , we can take any two out of e1 , e2 , e3 . For definiteness sake, let us take the
basis {e1 , e2 , e1 + e2 + e3 }. Similar reasoning applied to the span of {e4 , e5 , e6 } allows to take
{e4 − e6 , e5 − e6 , e6 }. Note that we could have taken {e4 − e6 , e5 − e6 , ej }, j = 4, 5 also as a
basis for the span of e4 , e5 , e6 .
We thus end up with a basis {e1 + e2 + e3 , e1 , e2 , e4 − e6 , e5 − e6 , e6 }. The new elements,
e1 , e2 , e6 that augment the basis of ker T are to be mapped bijectively to the chosen basis of
Im T . We shall define the linear map T by specifying its action on the above basis of R6 :

T e1 = v1 , T e2 = v2 , T e6 = v3 , T (u1 ) = T (e1 + e2 + e3 ) = 0, T (e4 − e6 ) = 0, T (e5 − e6 ) = 0.

To write the linear map in terms of the standard basis, we need to see what T does to
the elements of the standard basis. Already, we know the images of e1 , e2 , e6 . Hence we need
only find out T (e3 ), T (e4 ), T (e5 ). Since T (e1 + e2 + e3 ) = 0,

T e3 = −T e1 − T e2 = −v1 − v2 = −(f1 − f4 ) − (f2 − f4 ) = −f1 − f2 + 2f4 .

Since T (e4 − e6 ) = 0 = T (e5 − e6 ), we conclude that

T e4 = T e5 = T e6 = v3 = f3 − f4 .

Thus

T (x1 , . . . , x6 ) = x1 T e1 + · · · + x6 T e6
= x1 (f1 − f4 ) + x2 (f2 − f4 ) + x3 (−f1 − f2 + 2f4 ) + x4 (f3 − f4 )
+ x5 (f3 − f4 ) + x6 (f3 − f4 )
= (x1 − x3 )f1 + (x2 − x3 )f2 + (x4 + x5 + x6 )f3
+ (−x1 − x2 + 2x3 − x4 − x5 − x6 )f4
= (x1 − x3 , x2 − x3 , x4 + x5 + x6 , −x1 − x2 + 2x3 − x4 − x5 − x6 ).

It is trivial to see that ker T = U and Im T = V .

Let y := (y1 , . . . , y4 ) ∈ V . Note that y4 = −(y1 + y2 + y3 ). Equating T x = y, we see that


x1 = y1 + x3 , x2 = y2 + x3 and x4 + x5 + x6 = y3 . So, an obvious choice is let x3 = 0 and
x4 = x5 = x6 = y3 /3. Then T (y1 , y2 , 0, y3 /3, y3 /3, y3 /3) = y.

It is to be noted that such maps are far from being unique. Can you think of another
linear map different from the one given above?

Ex. 1. Let U := {x ∈ R2k : x1 + 3x3 + (2k − 1)x2k−1 = 0, x2 = x4 = · · · = x2k }. Construct a


linear map from R2k onto some Rl (l to be specified by you!) with U as its kernel.

You might also like