Basic Linear Algebra

Prof. Inder K. Rana

Department of Mathematics
Indian Institute of Technology - Bombay

Lecture – 02
Introduction II

Now let us look at slightly more complicated things namely if we are given more than one equation
in 3 variables, what could be possible solution of that.
(Refer Slide Time: 00:36)

So let us look at, which again geometrically it is very obvious.

(Video Start Time: 00:40 - Video End Time: 04:06)
There are 2 linear equations in 2 variables. There are 3 types of solutions there. Depending upon
the vertices in parallel, so there are no points of intersection and the system has no solution. And
if the lines are identical, the 2 perhaps coincide intersecting at an infinite number of points. The
system therefore has an infinite number of solutions. Likewise, systems of 3 linear equations and
3 variables can have a single unique solution, no solutions, or an infinite number of solutions
depending on ways in which the 3 planes are oriented.

Let us consider all possible ways in which 3 planes can intersect or not intersect. We will number
these planes 1, 2 and 3. One possibility is that no 2 planes in the system are parallel. Then all 3
planes intersect at only 1 point. In this case, the system will have 1 unique solution. The second
possibility is that 2 of the planes are not parallel and so their points of intersection form a line. If
the third plane intersects the other 2 along the same line, then the common points of intersection
for all 3 planes are represented by that one.

The system will therefore have an infinite number of solutions corresponding to every point on the
lock. The third plane is not necessarily have to be the same for both of the other points. If the third
point is that integral for one of the other 2 planes, then the points come into all 3 planes are still
represented by the same line of intersection and the system will have an infinite number of
solutions.
An infinite number of solutions will also exist if the system consists of 3 identical planes. In this
case, any point which lies on one plane is come into all 3 planes. The system will therefore have
an infinite number of solutions corresponding to every point on the plane. In addition, there are
several orientations of the plane before result in a system with no solution. Any time the system
has 2 distinct parallel planes, there can be no solution.

Since distinct parallel planes have no points in common regardless of the orientation of the third
point, there can be no point which lie on all 3 planes. The third plane can be distinct parallel to the
other 2 identical to one of the other 2 or not parallel to either of the other planes. Thus intersecting
third point.

Regardless of the orientation of the third plane since there are no points in common to all 3 planes,
the system has no solution. One additional configuration of the planes which will result in the
system with no solutions is that the planes are oriented so then there intersection points lie along
3 distinct parallel lines. Once again since there are no points in common to all 3 planes, the system
has no solution.
(Refer Slide Time: 04:07)

(Refer Slide Time: 04:13)

(Refer Slide Time: 04:20)

Right, so let us just summarize what we have done till now. We have looked at system of a linear
equation in 2 variables. And then we looked at 2 equations in 2 variables and then we looked at
the system of 3 equations in 3 variables and possible solutions. In either of this dimensions, either
the system has no solution or it has infinite number of solutions or a unique solution. How does
one go about analyzing when the number of variables increase?

There is no geometry available. What do we do about it? So what we will do is, try to convert these
geometric pictures into algebra and then try to see whether we can extend that algebra for more
number of variables.
(Refer Slide Time: 05:22)

So let us analyze this method of variable elimination method. So let us look at this example, simple
example and then we will try to formalize this. So example is 𝑥 + 3𝑦 + 5𝑧 = 0, 𝑥 + 2𝑦 + 3𝑧 = 1
and the third equation is 𝑥 − 𝑧 = 1. So I have labelled this equations as 𝐸1 , 𝐸2 , 𝐸3 . So the idea is
try to eliminate. Already in the third equation, right, 𝑦 is missing. So let us try to eliminate the
variable 𝑦 from the other 2 equations also, okay.

So that is the idea. So what we do is, we do these operations, okay. So here for example what we
have done is we have eliminated x from the equations. The first equation remains as it is. The
second equation what we have done is 𝐸3 − 𝐸1 and then operation is 𝐸2 − 𝐸1 . So we have just
taken a linear combinations of the equations. So the solutions will not change. The idea is whenever
you take a linear combination of any 2 or more of the equations, the solution does not change.

And now from these 2 equations, I can try to eliminate again one of the variables. So let us do that.
So 𝐸3 − 3 𝐸2 , so that operation is done.

So I get the system of equations which is equivalent to the earlier one because I have just taken
linear combinations, done nothing else, right. So solution of the last system should be same as
solution of the original system. So that is the idea. But in the last system, I get an equation 0 =
−2, that means that is not possible, that is absurd, right. So that means this system has got no
solution, right. So that is algebraically solving.
(Refer Slide Time: 07:41)

So this system has no solution.

Now consider this. The second system is equivalent to a given system. So solution of this should
be same as the solution of the original one. And here, I get 𝑥3 = 2, last equation. I put the value
of 𝑥3 in the previous equation, I get the value of 𝑥2 and I put these 2 values in the first equation, I
get the value of 𝑥1 .

So I eliminate and then substitute. So this method is called elimination and substitution method or
backwards substitution and that gives me the solution of the system. So this system has got a
unique solution, right. Earlier one, had no solution. This one has unique solution. And let us look
at another one, this one.
(Refer Slide Time: 09:13)
So there is only 1 equation 𝑥1 + 𝑥2 + 𝑥3 = 6 that because geometrically is a plane, right. How
many points are there in the plane? Infinite number of them. So this system, one equation in 3
variables will have infinite number of solutions possible. How I can write them? One possibility
is you can write 𝑥1 = 6 − 𝑥2 − 𝑥3 , right. So that means what? 𝑥1 can be determined in terms of
values of 𝑥2 and 𝑥3 and 𝑥2 and 𝑥3 are free to choose any values, right.

So I put different values for 𝑥2 and different value for 𝑥3 , I get different value for 𝑥1 . So for 𝑥2
and 𝑥3 can take infinite number of values, so that means 𝑥1 also has infinite number of values.
That means the system has got infinite number of solutions which we are verifying algebraically.
Geometrically it is plane, we know it is an infinite number of solutions. So 3 possibilities. I have
given you 3 examples where the system had no solution, system had exactly 1 solution and the
system had, okay, infinite number of solutions.
(Refer Slide Time: 10:32)
What is we observe in solving all these things? The solution of a system does not change whether
it is 2 variables 2 equations, 1 equation. Any 2 equations are interchanged, what will happen? If I
interchange any 2 equations, right, instead of saying this is line 1 and this is line 2 in the plane, I
am saying this is line 1 and this is line 2. I am just renaming them, right. The solution does not
change if I change the order of the equations, right.

An equation is multiplied by non-0 scalar. If I take a line, right and multiply it by a non-0 scalar
everywhere on the left hand side as well as right hand side, does the equation change? Equation
remains the same. Line remains the same, right. So solution does not change, right. So why non-
0? Because when I multiply by 0, then all the information is lost, 0=0, that is nothing, equation is
gone, right.

So multiplication by a non-0 scalar leaves the equation. So solution are unchanged. And the third
one is, one equation is added to another, that we had already seen, right. If one solution is there,
right and you add one equation to another scalar to multiple of that and now of course not 0, right,
then the solution does not change, right, that we saw. In the 2 variable thing, we saw that if you
add scalar multiple of 1 linear equation to another, the point of intersection does not change.

Only the inclination of the line changes, right. So these 3 are basic operations that do not change
the solution of a system of equations, right. So that is one observation. And then the other
observation is that the variables really do not play any role in all the computations. x1 remains the
x1, x2 remains the x2, x3 remains x3. It is only that constants which are in front of them or on the
right side of the equality, right.

They change when you do something with them, scalar multiple and do something, right. So we
can forget when we write. We can forget about these variables. Only we should keep track that
base is the coefficient of the first variable. This is the coefficient of the second variable. This is
coefficient of the third and so on, right. So variable do not play any role. So let us try to describe
this in an abstract study without observations.
(Refer Slide Time: 13:17)

So we will define formally an equation of the form 𝑎1 𝑥1 + 𝑎2 𝑥2 + ⋯ + 𝑎𝑛 𝑥𝑛 = 𝑏1 is called a

linear equation in 𝑛 variables 𝑥1 , 𝑥2 , … , 𝑥𝑛 . Number of variables is 𝑛, only 1 equation. Why it is
called linear? Because the powers of all the variables is 1. That is why it is called the linear equation
in n variables. So more than 1, so you can have a system of m equations in n variables.

So the first letter indicates the equation. Second indicates the variable, right. And for
computational purposes, we do not want 𝑥1 , 𝑥2 , … , 𝑥𝑛 coming into the picture, right. We do not
want them.
(Refer Slide Time: 14:46)
 c
So let us write, so before that let us just say a vector 𝑠1 , 𝑠2 , … , 𝑠𝑛 ∈ ℝ𝑛 . So when we say it is a
solution, if I put 𝑥1 = 𝑠1 , 𝑥2 = 𝑠2 , … , 𝑥𝑛 = 𝑠𝑛 in all the equations, then the left hand side is equal
to the right hand side, right. So that is called the solution. Though the idea is how do we solve such
a system, that is what we want to analyze?
(Refer Slide Time: 15:41)

So let us write that. Variables are 𝑥1 , 𝑥2 , … , 𝑥𝑛 equality and the right hand side something appears,
right.
So return this way, all the data about the system is captured. We have fixed the position of the
variables. This is captured.
(Refer Slide Time: 16:25)

So this motivates, so a rectangular array of 𝑛 equations in 𝑛 variables. So basically what we are

saying is to represent a system of a linear equations, a rectangular array of numbers becomes
important, right. So we start defining, because this is going to come again and again. It motivates
a definition in mathematics, namely a rectangular array of numbers, they could be real or complex.

In fact, they could be more general but we will be concerned only with constants which are real or
complex, okay. So a rectangular array in which there are m rows, first row, second row, third row,
m rows are there. And each row has got a 𝑛 column, right. So this is called a matrix, okay. Why
matrix is important you can ask?

The important thing is when we do computations, you are not going to do it humanly. We are
going to put it on a computer, all the computations, right. All multiplications, divisions, everything
and a computer or a machine cannot store variables. We can only store scalars, right. So we can
ask the machine to store the data in this format, right. Even machine will not know what is the row
1 or row 2 or row 3 or row 𝑚, right. So what you do it is write all if there is a rectangular array of
m numbers in rectangular array, 𝑚 × 𝑛.

How many total numbers are there? 𝑚 × 𝑛, right. So the computer will store it as 𝑚 × 𝑛 numbers
So the basic idea is matrices written this way capture the computational aspect of linear equations,
okay. So this is called a matrix of order 𝑚 × 𝑛. When 𝑚 = 𝑛, we will call this as a square matrix.
(Refer Slide Time: 19:04)

So if you are in the 𝑖th row and 𝑗th column, so that position is called the 𝑖𝑗th term or entry of the
matrix. So we will start using this term. We got a matrix whose m rows, n columns, ijth entry is so
and so, right. So we will start using that language. So sometimes in short you write this whole, if
you are not really interested in numbers.
(Refer Slide Time: 20:11)

So we have defined a new quantity which involves numbers. So what are the possibilities? What
kind of operations we can do? So this is very common in mathematics. You define a quantity. You
collect all such quantities and try to do operations on these quantities. For example, you must have
studied functions, right, real value functions. Then you can think a real value functions, defined
on the interval.

We can add functions. We can multiply functions. We can scale or multiply a function. You can
compose functions. So in the class of all functions, you can do these various operations. You can
add, you can subtract, you can scale or multiply, you can multiply functions or you can compose
functions. Similarly, we have got collection of matrices, right. So what do we want to do?

First of all, given 2 matrices of the same order, we say they are equal so we are defining equality
of matrices. The order is same. So natural thing to say whether they are equal means what, each
entry should be equal. So 𝑖𝑗th entry of 1 should be equal to 𝑖𝑗th entry of other, right, the
corresponding entries must be equal. So that is equality of matrices.