Basic Linear Algebra

Prof. Inder K. Rana

Department of Mathematics
Indian Institute of Technology - Bombay

Lecture – 05
Systems of Linear Equations II

(Refer Slide Time: 00:26)

Let us look at the matrix 0 0 0 0. Is it in the row echelon form? We are looking only at the non-0
ones, so we are not really bothered about it, okay. You can say this is not in the row echelon form.
It cannot be brought because it is non, there is no non-0 entry at all. Let us look at this 1 2 and 5.
Is that in the row echelon form? Yes. It is. There is only 1 row which is non-0, so fine. What about
this one? That also is, okay.
(Refer Slide Time: 00:58)
Let us look at some more. For example, let us look at this one. 0 1 2 first row. 1 0 2 then second
row, right. Here the first non-0 entry in the first row is coming at 1 but in the next row, it is coming
at, first place itself. Whereas in the second column where it is in the first column, so this is not in
the row echelon form. If I wanted to bring it in the row echelon form, I will interchange these 2
rows.

(Refer Slide Time: 04:07)

(Refer Slide Time: 05:56)

So let us look at, so this is a theorem which can be written down, the proof can be written down.
Actually when we do the changing a matrix to the row echelon form, we will see that the proof
automatically will be a part of it. There sometimes, you say there exists, you actually construct
that thing and that shows that it exists also, right. So there are some proofs in mathematics which
are purely existential proofs where you cannot show that object but you can prove it exists, right.

There are other proofs in mathematics where you actually construct and show that it exists. The
difference is very my pet quote saying that the god exists, right. Some people believe god exists
because of some reasons. But you cannot produce god in front of you, right. So there are existential
proofs which purely say, right. You might have seen in calculus that a criteria for convergence of
a limit.

Every equations sequence has a limit. It does not tell you where it is. It says it exists, right. But
definition of convergence says you can locate the point which is the limit. So that is the difference
between existence and proving actually it exists, okay. You will come across these things in many
parts of mathematics

There is a typo here. Every matrix, right, of any order 𝑚 × 𝑛, there is no relation between 𝑚 and
𝑛, okay. There is a typo. It should be 𝑚, 𝑚 × 𝑛 matrix can be reduced to a row echelon form. The
word a is used indicating that row can be more than 1, row echelon form of a matrix, using purely
elementary row operations, okay. So the row echelon form we will shorten it as REF. Row
operations we will write as row operations. Is the theorem, we will see how the proof also goes as
we do something, okay.
(Refer Slide Time: 08:11)

Why the word echelon comes? If you look at the English dictionary, what is echelon? So go back
and look at English dictionary. What does the word echelon means? From where does this word
suddenly drops in, in mathematics? Echelon means a particular arrangement formation of
something. For example when the planes fly, they fly in a particular form. It is called the echelon
form of the planes.

So one is behind the other, right. They do not, one position is slightly behind. So the position where
the non-0 entry is coming is on the right side, that is the formation. So echelon word basically
means formation, particular formation of something, okay. I just brought this picture now.
Hopefully the echelon word will stay in your mind. It is formation of one going behind or staircase
if you want to keep in mind that kind of a thing, okay.
(Refer Slide Time: 09:15)
So let us look at an example how does one do that. All should be obtained only using row,
elementary row operations. That is the only condition. Those are the only tools given to you, right.
Using those tools, you have to transform the given system into a new system, that is all, okay. So
there is row echelon form. Let us go a step further.
(Refer Slide Time: 14:01)

Let us look at one more example, okay.

So you can say that the row echelon form means that there is a 𝑟, so the bottom 𝑟 rows are 0. In
the non-0 row, the non-0 entry is pivot, right. The pivots come at in the column which are
increasing order. That is all, right. You can understand that way also, right. So let us pick up one
of them which is not say.
(Refer Slide Time: 17:52)

Which one? a) is not, right.

(Refer Slide Time: 18:18)

Now here is a proof of that theorem or a process also. You see the point is when we are saying it
is a calculation of numbers only, the variables are not important. That means the numbers, the
values of the entries of the matrix, we are going to multiply by something, add 1 row to another
and do something. And here 2/3, 3/2, 4/5, you can do humanly. You can use calculators and do it.

But in applications, these things, the number of equations, you will be surprised can run into
thousands. For example this weather prediction, how do they predict weather? How do they make
model for economy? There are lot of equations coming. Lot of factors which are effecting the
equations coming. And more often than not, to solve these equations one make them linear that is
called approximation.

So you linearize the equations, the model, you have a model in which lot of equations are coming
which are not linear. You linearize it. So you get an approximation kind of a thing, right. Because
when a system is linear, it is easier to solve. We have methods of solving them. So how does one
solve a system of linear equations is very important thing for applications point of view also.

And the idea is that even if the matrix is order or 1000*1000, 1000*2000, we can put it on a
computer and we can ask the computer or a machine to do the calculations for us and make it in
the row echelon form. But the question is, how does computer knows what to do, right. They only
may have numbers. So that is why the matrices coming to the picture. The variables are gone. Only
the entries of the matrices and that 𝑏1 , 𝑏2 , … , 𝑏𝑚 , the column that is important.

So and that we can feed a data to the computer, right. When you are saying interchange, we will
see later on how those operations can be done by computers. But we have to give a definite method
of doing it.

There is no definite way of saying that okay this is the only way of changing a matrix to row
echelon form. So can we give a definite way by which we will proceed so that the computer knows
we know, everybody knows that this is the way we have done it. So that method is also the proof
of this theorem that every matrix can be reduced to row echelon form.