Prapanch Nair, AM, IITD

Engineering Mathematics
& Numerical Methods
Lecture 1

1. Linear Algebra
1.1. Matrix theory
matrix

is the element of the matrix (complex or real) at the row column.

Row matrix is , column matrix is .

Square matrix is of size .
Trace sum of diagonals.
and are equal if
is a zero matrix if .

Matrices are conformable tor addition (sub) if they are of the same size.
The following relations hold with respect to matrix additions:

Matrix multiplication
is and is
is with elements

If

1 / 13
Conformable for multiplication if colums of Rows of .
The following relations hold:

Types of matrices
Identity matrix
Upper triangular if
Similarly lower triangular if

Inverse matrix: , then is Non-singular

and

Transpose:
The following results apply

A matrix is symmetric if , skew symmetric if

,

Determinant:
denoted as
The following hold:

Minor is det of matrix contructed by removing row column.

The cofactor matrix is defined as follows:

2 / 13
If then
If exists then

Linear system

It is non singular

(How expensive is this computation?)

What can we do knowing matrices?
minimize
Factor the matrix .
Find and
Is in column space of ? is there a solution?
so Ax kups same direction
so is just

Columnspace of a matrix
How do you usually multiply?

row by column.
What if we multiply column by rows?

3 / 13
linear combination of and .
i.e, multiply by scalars and add.
Thus Ax is a linear combination of colums of . All possible combinations of the columns Column space of .
All combinations of produces the subset of space, which is a plane!

is in the colvenn spare of exactly when has a sotection

Let's define vector space properly now.

Lecture 2
1.2. Vector space
A vector space is a triple of an Abelian Group, a Field and an operation called scalar multiplication:
Abelian Group :
Group: is associative binary operation and a set with identity and inverse of 'all elements' of G w.r.t.
to . If group is commutative, then it is Abelian.
eg: with is group. with + is group.
Properties:
1. Identity is unique.
2. Inverse is unique for each element.
3. Commutative w.r.t. .
Field : A field F is a triple (D , +, ⋅ ) consisting of a set D and two binary operations + and ⋅ such
that:
(a) D with the operation + is an Abelian group.
(b) The operation ⋅ is associative.
(c) D contains an identity element, denoted by 1, with respect to the operation ⋅ , i.e.,
1 ⋅ a = a ⋅1 = a
for all .
(d) The operations + and ⋅ satisfy the distributive axioms
a ⋅ (b + c) = a ⋅ b + a ⋅ c
(b + c) ⋅ a = b ⋅ a + c ⋅ a
(e)The operation ⋅ is commutative.
(f)If a , b, c are any elements of D with c ≠ 0, then
a ⋅ c = b⋅ c ⇒ a = b
(g) containing more than one element, and such that any
element has an inverse with respect to multiplication
function f: called scalar multiplication such that

for all , μ ∈ F and all u, v ∈ V . A vector is an element of a vector space. The notation ( ) for a vector space
will be shortened to simply V . The first law is usually called the associative law for scalar multiplication, while the
second and third equations are distributive laws,
the second for scalar addition and the third for vector addition.

4 / 13
 is basically the scalar multiplication

Henceforth we will call as

Properties:
(a) has the binary operation called which satisfies:
i)
ii)
iii)
ii) for every
b) , scalar can combine with s.t
i)
ii)
iii)
iv) and

if , real vector space.

But in general complex.

Examples:
1.
and and
iff for all
also

and (scalar multi..)

is cartesian product of the complex numbers .
2. Set of all complex matrices is a vector space w.r.t + , by scalar is a V.S.
3. be a V.S whose are functions defined on a set with values in . Thus if then
and . If is another vector of it then:
equality:
zero vector is a tunction whose values are 0 for any .

e. g.
4. P , a set of all polynomials of degree

The set forms a V.S.

5. Set of of all with + ,
6. 0 vector of any V.S forms a V.S itself
Important is not a binary operation

5 / 13
Thorem
or .
Proof:
assertions needed
a)
b)
c) or .
a) take. in Axiom b2 for V.S. then

using and

proving a.
b) Set in axiom b3

Hence proved.
c) assume
if , from a is true.
if , then we must show must be 0.

Theorem:

Let , replace with in .

Theorem:

Lecture 3
1.3. Linear Independence, Dimension, Basis
Linear Dependence.
Set of vectors in a V.S. is linearly dependent if there exists not all
zeros s.t.

Which means one vector can be expressed as linear combination of all others.
that means if then

Example:

6 / 13
1.3.1. Theorem
If are linearly dependent, then every finite set containing them is linearly dependent.
Every set with a 0 is linearly dependent.
When not lin dep. They are lineary independent.
A set of vectors is lin independent if (25) (the only way 25
can be written).
Consider from (what is this? cartesian product) Thuy are lin. indep.

1.3.2. Theorem:
Every non-empty subset of LI set is LI.
A LI set is said to be maximal if it is not a proper subset of any other LI set.
V.S with a finite maximal finite dimensional v.s.

1.3.3. Thorem:
Any two maximal LI sets of a finite dim VS must contain same number of vectors.

Proof:
and are both maximal and sets of

we must show .
Lets say . and that .
Because is maximal, the sets are LD. Then

coefficients are not all

To show 's cannot be . unless .
The coefficients are all (each) non zero because if for any , then.

contradicting the assumption that is LI.

So because 's are all non zero we can write

7 / 13
where for . We can claim that 1 equations can be inverted S.t.
are written as L.C of

Inversion is possible if for is non singular

Now we claim that the first equations in the above system can be inverted in such a way that the vectors
are given by linear combinations of the vectors . Indeed, inversion is possible if the
coefficient matrix for is nonsingular. But this is clearly the case, for if that matrix were
singular, there would be nontrivial solutions to the linear system

Then from (9.3) and (9.4) we would have

contradicting the assumption that set , being a subset of the linearly independent set
, is linearly independent.

We'll chack that in a bit.

If is non singular then the inversion gives

where for . we can substitute (26) into the remaining eq.

meaning are lin comb. of .

thus is impossible

Corollary
1.3.4. Theorem:
Let be maximal set in and suppose be given by (26) . Then is
also maximal, iff is non-singular.

If In particular, it for but , then

is maximal, iff in the expansion of in term of in is
non zero.

8 / 13
Lecture 4
Number of a s is a property dimension
Any maximal of is a basis.
As soon as one basis is known, all basescan be computed from thorem 9.5
1. Canonical basis. for , or standard basis.
2. is a VS with clements from , thm

3- 1 and torm a basis for vector space of complex numbers over a field of .
9.6: It is a basis tor , then every veitor in has the representalion

T9-7: The scalars are anique.

Proot:
Assume non uniquenes.

Subtrant

For of we need to . The coefficients in 9.6 are components of V. w.r.t.

Examples of basis
1. V.s has a standard basis

2. in Field of

3.

T9.8: is a bahis and is an set , then its possible to exchang wi A set of

vectors of are such that overy

is a generating set.
9.10. Equivalent conditions:
i) is maximal LI

9 / 13
ii) is generating set.
iii) is minimal generating set.
is called the basis of the vector space.

1.4. Inner Product Spaces

Inner product spaces:
Length or magnitide.
Inner product is a function with following properties.

1. ; (conjugate)
2. (hicar on first argument)
3.
4. and
and

(1) ensures. is read

(L) ensuris is tre difinite.

Inner product space is a Vs with and inver product.

from .
c) 1)
2)
3)
4) and

Rial IPS is III.

Theorem 1. (Schwarz inequality)

Proof:
Trivial when or .
Construct and use .
non- negative norm

10 / 13
A generating said is said to span

ve square root gives Shwarz inequality

\begin{theorem}[Triangle inequality]

for any two vectors.

\end{theorem}
Proof :

use shuvary

Take ve sq. root:

Angle:

The vectors are orthugonal when or .

Distance

Coordinate/Components

for the hasis

Orthonormal set of vectors are such that

11 / 13
where kronecker delta.

Theorem 2. Orthonormal set is LI.

Proof :

Take dot product with

which is a contradiction
A complet set is a basis
Orthonormal basis.

1.4.1. Gram - Schmidt orthogonalization

Theorem 3. Given of and Inner product space , then there exists basis s. t.
and generate the same subspace. of for each

Proof: is obtained also by construction.

i) set of orthogonal vectors are constructed and then normalized.
Let orthogonal but not unit vectors.
we can construct these from a given basis
Let

choose so that

since .
is non jero because , and are .

and obtaned from and

of and required is non zero.

Repeat until then .
It is easy to see that

generat same subspace for each .

12 / 13
Orthogonal complement of 7 U

Gram schnidt.
example.

Check.

13 / 13