Sie sind auf Seite 1von 69

Contents

Useful resources 4

Vectors 5
Vector space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Linear independence and linear dependence . . . . . . . . . . . . . . . . . . . . . . . 7
Basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Dimension of a vector space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Norms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Norms over a real and a complex space . . . . . . . . . . . . . . . . . . . . . . . 9
Norms over polynomial spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Scalar (or dot) product of vectors over a real and a complex space . . . . . . . . . . . 10
Scalar (or dot) product of vectors over a polynomial space . . . . . . . . . . . . . . . 10

Matrices 11
Spectial Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
I. Square Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
II. Rectangular Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Equality, Addition, Multiplication by a scalar . . . . . . . . . . . . . . . . . . . . . . 12
Matrix multiplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Matrix Multiplication Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Invertible Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Matrix transposition and related matrices . . . . . . . . . . . . . . . . . . . . . . . . 14
Determinant, minor and rank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Linear systems 17
Solution of a linear system and its geometrical interpretation . . . . . . . . . . . . . . 17
Finding a solutions of systems of inhomogeneous linear equations . . . . . . . . . . . 19
Gauss elimination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Cramer’s rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Solutions of systems of homogeneous linear equations . . . . . . . . . . . . . . . 20
Remarques on numerical solution of systems of linear equations . . . . . . . . . . . . 20

Exercises with matrices and linear systems 21

Derivatives 22
Limit of a function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Rules to calculate derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Interpretation of derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Table of Key Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

1
Derivatives of functions of multiple variables 28
Partial derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Chain rule for the function of multiple variables. Jacobian matrix . . . . . . . . . . . 29
Directional derivative . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

Taylor’s theorem, Taylor’s polynomial, Taylor and MacLaurin Series 33


Taylor’s theorem in one real variable . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Taylor’s and MacLauren series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
D’Alambert ratio test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

Indefinite integral or anti-derivative or primitive 37


Substitution rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
Integration by parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Recomendations for the integration by parts . . . . . . . . . . . . . . . . . . . . . . . 39
Recommended substitutions for indefinite integral . . . . . . . . . . . . . . . . . . . . 40

Definite integral 42
Defined integral: properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Fundamental theorem of calculus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Substitution rule for definite integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

Multiple Integrals 45
Product under the integral sign and product of the integrals . . . . . . . . . . . . . . 46
Double Integrals Over General Regions . . . . . . . . . . . . . . . . . . . . . . . . . 46
Gaussian Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

Integrals in physics: line integral, surface integral 51


Line integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Parametric presentations for some curves two dimensions . . . . . . . . . . . . . . . . 52
Line integral with respect to the coordinates . . . . . . . . . . . . . . . . . . . . . . . 54
Vector fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Line integral of Vector Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Green theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

Surface integrals 59
Divergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Operator of Laplace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Vector form of Green’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Surface integral and the Divergence Theorem . . . . . . . . . . . . . . . . . . . . . . 60

2
Differential equations 62
Ordinary differential equations of first order: first order ODE . . . . . . . . . . . . . . 63
Separation of variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Substitutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Exact first order ordinary differential equation . . . . . . . . . . . . . . . . . . 66
Inexact equations: integrating factors . . . . . . . . . . . . . . . . . . . . . . . . 67
Variation of a constant parameter . . . . . . . . . . . . . . . . . . . . . . . . . . 69

3
Useful resources
Theory:
On calculus (derivatives, integration, etc):
http://tutorial.math.lamar.edu/

Wikipedia on a concrete subject

Exercises:
On calculus (derivatives, integration, etc) the same resource as the theory:
http://tutorial.math.lamar.edu/

On matrices:
https://matrix.reshish.com/

Everything (little explanations)


https://www.wolframalpha.com/examples/Math.html

Plot of graphic
https://graphsketch.com/

4
Vectors
Vector space
Definition 1: Closed set:
The set is closed under a certain operation if and only if the result of the operation belongs to
the set.
Defitions 2: Vector space:
A vector space V is a collection of objects with two operations: (vector) addition and multipli-
cation by a scalar (number). This collection (vector space) is closed under these two operations.
The two operation should satisfy the following axioms ∀~x, ~y , ~z ∈ V and α, β ∈ R

1.(α + β)~x = α~x + β~x


2.α(β~x) = (αβ)~x
3.~x + ~y = ~y + ~x
4.~x + (~y + ~z) = (~x + ~y ) + ~z
5.α(~x + ~y ) = α~x + α~y
~ ∈ V 3 ~0 + ~x = ~x; ~0 is usually called the origin
6.∃O
7.0~x = 0
(1)
Examples of vector space

Example 1: Linear vector space R2 = {(a1 , a2 )|a1 , a2 ∈ R}


Remainder: Graphical representation of the vectors in two-dimensional space R2 : the pair a1 , a2
is interpreted using (x, y) coordinate plane (fig.1)

Figure 1: Graphical representation of a vector from a linear vector space R2 using (x, y) coor-
dinate plane

How to define the addition of vectors and their multiplication by a scalar in a R2 ? These
definitions are up to us providing they satisfy axioms formulated at the beginnig
of the chapter! For example, let define the addition of two vectors (a1 , a2 ) and (b1 , b2 ) as :
(a1 , a2 ) + (b1 , b2 ) = (a1 + b1 , a2 + b2 ) (2)
5
A graphical representation of the ”just defined” vector addition in R2 corresponds to a so-called
parallelogram rule (fig.2)

Figure 2: Graphical representation of the addition of vectors in a linear vector space R2 using
(x, y) coordinate plane

Let define multiplication of a vector (a1 , a2 ) by a scalar α ∈ R as

α(a1 , a2 ) = (αa1 , αb1 ) (3)

Example 2: Linear vector space Rn = {(a1 , a2 , ..., an )|a1 , a2 , ..., an ∈ R},


Remainder: Graphical representation of the vectors in three-dimensional space R3 : is similar
to that in R2 and is interpreted using (x, y, z) coordinates.

Example 3: C2 and Cn respectively to R2 and Rn where the underlying field is C, the complex
numbers.

Example 4: Linear vector polynomial space of all polynomials of degree n (including those
whose degree is less than n) Pn
( n
)
X
Pn = aj xj |a0 , a1 , · · · , an ∈ R (4)
j=0

Graphical Interpretation: In a linear vector polynomial space P1 a vector can be presented


using the x-axis for the coefficients before x0 and y-axis for coefficient a1 (fig.3), so the dimen-
sions of P1 is two. Can we define the addition of vectors and their multiplication by a scalar in
a R2 ? Yes, these definitions are up to us providing they satisfy axioms formulated at
the beginning of the chapter! For example, let define the addition of two vectors in P n as
n
X n
X n
X
j j
aj x + bj x = (aj + bj ) xj (5)
j=0 j=0 j=0

6
Figure 3: Graphical representation of a vector from a linear vector polynomial space P n using
(x, y) coordinate plane

and multiplication by a scalar α as


n
X n
X
j
α aj x = (αaj ) xj (6)
j=0 j=0

It should be verified that introduced operation satisfy the axioms.

Linear independence and linear dependence


Definition We say that
a1~v1 + a2~v2 + · · · + ak~vk (7)

is a linear combination of the vectors ~v1 , ~v2 , · · · , ~vk .


Definition Let us choose a set of vectors S = {~v1 , ~v2 , ..., ~vk } ⊂ V (S is a vector space, subset
of V). The set S is said to be linearly independent if for any scalars a1 , a2 , · · · , ak of which at
least one coefficient is not zero a linear combination of these vectors is not zero:

a1~v1 + a2~v2 + · · · + ak~vk 6= 0

Otherwise, the S is said to be linearly dependent.

Exercise:
Consider a linear vector space R2 = {(a1 , a2 )|a1 , a2 ∈ R}.

Question 1 if vectors ~x = (−1, 5) and ~y = (2, −3) are linearly dependent.


Question 2 Can you propose (other) two linearly independent vectors?
Question 3 How many pairs of linearly independent vectors can you propose for a given space?
Question 4 How many linearly independent vectors can you propose for the given linear vector
space?

7
From the given definition it follows that if a vector space S = {~v1 , ~v2 , ..., ~vk } ⊂ V, is linearly
dependent, then one member of this set can be expressed as a linear combination of the others.
Indeed, if aj 6= 0, taking into account that

a1~v1 + · · · + aj ~vj + · · · + ak~vk = 0

we can solve this equation for ~vj :

1
~vj = − [a1~v1 + · · · + ak~vk ]
aj

Basis
Definition. A basis is a (finite or infinite) set S = {~vi }i∈N ⊂ V of vectors, that spans the
whole vector space V and is linearly independent. ”Spanning the whole space” means that any
vector ~u ∈ V can be expressed as a linear combination of the basis elements:

~u = a1~v1 + a2~v2 + · · · + an~vn

where the ak are scalars, called the coordinates (or the components) of the vector ~u with respect
to the basis S, and ~vik , (k = 1, ..., n) elements of S.
Note that if S = {~vi }i∈N ∈ V is a basis of V, then every vector ~u ∈ V has a unique represen-
tation in S. Indeed, suppose that a second representation is possible, so:

~u = a1~v1 + a2~v2 + · · · + an~vn ai 6= 0


~u = b1~v1 + b2~v2 + · · · + bn~vn bi 6= 0 (8)

then, using the axioms for the addition operation in a vector space, make a difference of these
two equality for the ~u:

0 = (a1 − b1 )~v1 + (a2 − b2 )~v2 + · · · + (an − bn )~vn

Since, by definition, the set of vectors {~vi }i∈N is linearly independent, it follows that ai = bi .

Dimension of a vector space


One of the most remarkable features of vector spaces is the notion of dimension. A Basis
Theorem says that if a linear vector space S = {~v1 , · · · , ~vk } is a basis for a vector space V,
then every basis for V has k elements. Using this theorem (without proof), we introduce the
following definiton.
Definition of a dimension of a linear vector space
The dimension of a vector space V is the (unique) number of vectors in a basis of V. We write
dim(V) for the dimension.

8
Examples:
1. dim(Rn ) = n
2. dim(P n ) = n + 1 (see the following example)
3. Consider a linear vector space of polynomials of degree n P n = {an xn + an−1 xn−1 + · · · +
a1 x 1 + a0 } .
Prove that the powers x0 = 1, x2 , · · · xn are linearly independent and since they span the P n ,
they form a basis of P n .
Proof demonstration in the class
Exercise: Prove the statement made above that dim(P n ) = n + 1
Exercise:
Let vector space M equal to all rectangular arrays of two rows and three columns with real
entries. Find a basis for M, and find the dimension of M. Note
(" # )
a b c
M= a, b, c, d, e, f ∈ R (9)
d e f

Norms
Norms are a way of putting a measure of distance on vector spaces. First, we shall present
norms over R and C and then norms on polynomials vector space.

Norms over a real and a complex space

Definition Let V be a linear vector space and suppose that a norm on V, defined hereafter as
|| · || is a function from V to the nonnegative reals for which the follwing fulfils:

(i) kxk ≥ 0 ∀x ∈ V and kxk = 0 if and only if x = 0 (10)


(ii) kαxk = |α| kxk ∀α ∈ C, R and x ∈ V (11)
(iii) kx + yk ≤ kxk + kyk ∀x, y ∈ V The Triangle inequality (12)

Examples. Let V = Rn ( or Cn ).We can define for elements ~x = (x1 , · · · , xn ) of V the following
norms:

(a) kxk2 = (|x1 |2 + |x2 |2 + · · · + |xn |2 )1/2 Euclidean norm (13)


(b) kxk1 = (|x1 | + |x2 | + · · · + |xn |) l1 -norm (14)
(c) kxk∞ = max |xi | l∞ -norm (15)
1≤i≤n

It can be proved that all norms defined by eqs.13,14 and 15 correspond to the definition of the
norm given above. Yet, how to deal with 3 different norms?
Definition Let k·ka and k·kb be two vector norms on Cn . We say that these norms are equiv-
alent if there are positive constants m, M such that ∀x ∈ Cn m kxka ≤ kxkb ≤ M kxka .

9
Theorem All norms are equivalent.
Exercise: Take as an example the linear vector space R2 .
Question 1: Can you interpret graphically all given norms?
Question 2: What means that all norm are equivalent if you compare two vectors from R2 ?

Norms over polynomial spaces

Polynomial spaces, as we have considered earlier, can be given norms as well. Yet, sometimes it
is difficult to consider all the values of the independent variable. In many cases a domain of the
polynomials has to be restricted. Let us introduce a notation P k (a, b) = P k for the polynomials
whose independent variable is restricted to a domain [a, b].
Now, let us define norms accepting that a polynomial from the P k (a, b) can be considered as a
function p(x):
Z b 1/2
2
(a) kp(x)k2 = |p(x)| dx
a
Z b
(b) kp(x)k1 = |p(x)|dx
a
(c) kp(x)k∞ = max |p(x)|
a≤x≤b

Scalar (or dot) product of vectors over a real and a complex space
Definition. If ~xquadandquad~y are vectors of the same length defined in Rn (or Cn ) , with
~x = (x1 , · · · , xn ), ~y = (y1 , · · · yn ), then the scalar product (or dot product) of ~x and ~y is given
by the formula:
n
X
~x · ~y = h~x, ~y i = xi y i (16)
1

Example. Let ~x = (1, 0, 3, 1) and ~y = (0, 2, 1, 2) then ~x · ~y = · · · (solved in the class).

Scalar (or dot) product of vectors over a polynomial space


Similar to the case when we introduced a norm over polynomial space, we will restrict poly-
nomials to an interval [a, b]. Then often the scalar product over a polynomial space is defined
as
Z b
f (x) · g(x) = hf (x), g(x)i = f (x)g(x)W (x)dx (17)
a

where W (x) is a weight function.

10
Matrices
Definition
Matrix is a collection of objects (such as numbers, polynomials, etc) ordered in rows and
columns. The size of the array is written as m × n, where m is number of rows and n is
number of columns.

Examples:
" #
a b c
A2,3 =
d e f
 
x2 + y 2 + 1 5xy 7
B3,3 = 11 8x + y 2 −3xy + 4x2 
 

4 + 5y − x2 + 3xy 8y + 1 −9xy + 6.5


Notation

columns
 
a11 a12 · · · a1n
 
 a21 a22 · · · a2n 
Am,n =
· · · · · ·
 rows
 ··· 

am1 am2 · · · amn
For a matrix given above:
· uppercase A denotes a matrix
· lowercase a denotes an entry (an element) of a matrix
· both, an indexed uppercase Aij and lowercase aij denotes an entry ij of the matrix

Notation
The set of all m × n matrices is denoted by Mm,n (or Mm,n (F ) where F is a field, it can be R
or C or polynomials, etc)
In a case where m = n we write M n (Mn (F )) to denote the matrices of size n × n.

Spectial Matrices
I. Square Matrices

If m = n, the matrix is called square. For a square matrix the following cases can be considered

1. Diagonal matrices
If aij = 0 for i 6= j and aii 6= 0, a matrix A is called diagonal
A diagonal matrix A may be denoted by diag(d1 , d2 , · · · , dn ) where aii = di

11
2. Identity matrix
The diagonal matrix diag(1, 1, · · · , 1)is called the identity matrix and is usually denoted by In
(cf illustration):  
1 0 ··· 0
0 1 · · · 0
 
In =  ..

0

0 . 0

0 0 ··· 1
The identity matrix can be simply denoted I if n is supposed to be known.

2. Lower triangular matrix


A square matrix L is said to be lower triangular if

lij = 0 for i < j. (18)

3. Upper triangular matrix


A square matrix U is said to be upper triangular if

uij = 0 for i > j. (19)

II. Rectangular Matrices

Definitions
A rectangular matrix A is called nonnegative if aij ≥ 0∀i, j.
A rectangular matrix A is called positive if aij > 0∀i, j.

Equality, Addition, Multiplication by a scalar


Definition Two matrices A and B are equal if and only if they have the same size and

aij = bij ∀ i, j

Definition If A is any matrix and α ∈ R (or α ∈ C) then the scalar multiplication is defined
as

B = αA with
bij = αaij ∀ i, j

i.e. all entries of a matrix is multiplied with a scalar.


Definition If A and B are matrices of the same size then their sum is defined as C = A + B
with

cij = aij + bij ∀ i, j

12
A theorem says that the defined operations of additions of matrices and multiplication by a
scalar satisfy the rules of (i) commutativity, (ii) associativity, (iii) distributivity of a scalar and
also to:
4. The equality α(βA) = αβA
5. A sum with a zero matrix (a matrix full of all zeros) for which A + 0 = A and α0 = 0.
6. (α + β)A = αA + βA
7. 0A = 0
8. α0 = 0

This theorem can be proved using the properties of scalars and of the field (sets) to which
element of the matrices belongs

Matrix multiplication
Definition
If A is m × n and B is n × p. C = AB is the m × p matrix with entries defined as

Pn
cij = k=1 aik bkj for 1 ≤ i ≤ m and 1 ≤ j < p

Another presentation of the matrix multiplication is the following. If A is m × n and B is n × p.


Let ri (A) denotes the vector with with entries given by ith row of A and let bj (B) denote the
vector with entries given by the j t h row of B. The product C = AB is the m × p matrix with
entries defined by scalar products of the chosen scalar:

cij = hri (A), bj (B)i

Matrix Multiplication Rules

1. If AB exists, it may not happen that BA exists. If both exists, generally AB 6= BA


2. if A is square matrix, we define

A1 = A, A2 = AA, A3 = AAA
An = An−1 A = A · · · A (n f actors)

3. Let I = diag(1, · · · , 1) then for Am,n :

AIn = A, and Im A = A

4. Note: if AB = BA it does not follow that A = B


Matrix Multiplication Rules. Assume A,B, and C are matrices for which all products
below make sense. Then
13
(1) A(BC) = (AB)C
(2) A(B ± C) = AB ± AC and(A ± B)C = AC ± BC
(3) c(AB) = (cA)B
(5) A0 = 0 and 0B = 0
(6) if A square, then Ar As = As Ar ∀ r, s ≥ 1

Invertible Matrix

Let A ∈ Mn (F ). The matrix A is said to be invertible if there is a matrix B ∈ Mn (F ) such


that

AB = BA = I

In this case B is called the inverse of A, and the notation for the inverse is A−1 .
Example. Let
" #
1 3
A=
−1 2

Find A−1 . Solutions: using the definition of the inverse matrix, compose a system of linear
equations. Solve it. You should get the answer:
 
2 3 " #
−  1 2 −3
A−1 =  51 15  =

5 1 1
5 5
Remark: A square matrix is NOT necessarily invertible. A matrix A should satisfy a certain
condition to be reversible, see the section of Linear systems.

Matrix transposition and related matrices


Definition of a matrix transpose
The transpose of Am,n , denoted by AT or Atr or A0 , is the n × m matrix with entries defined
as

(A)Tij = Aji

In words: The rows of A become the columns of AT , taken in the same order.

Notations
Let us denote as A∗ a matrix whose entries are complex conjugate to the entries of A: aij = a∗ij
Clearly, (A∗ )∗ = A

Properties of transpose For matrices A, B and scalar c we have the following properties of
transpose:
14
T
1. AT = A
2. (A + B)T = AT + BT
3. (AB)T = BT AT
4. (cA)T = cAT
∗
5. (A∗ )T = AT

Definitions for special transposed matrices


Symmetric matrix: A square matrix whose transpose is equal to itself is called a symmetric
matrix; that is, A is symmetric if

AT = A

Anti-symmetric or Skew-symmetric matrix: A square matrix whose transpose is equal


to its negative is called a skew-symmetric matrix; that is, A is skew-symmetric if

AT = −A

Note that if A is antisymmetric (skew-symmetric), then A is a square matrix and aii = 0,


for i = 1, ..., n.
Indeed: if A ∈ Mm,n (F ), then AT ∈ Mn,m (F ). So, if AT = A we must have m = n. Also
aii = −aii for i = 1, · · · , n that means that aii = 0 ∀ i.

Determinant, minor and rank


Definition The determinant of a matrix A is denoted det(A), det A or |A|. The following
formulae is applied in case of 2 × 2 matrix:

a b
|A| = = ad − bc

c d

A formulae that can be used in case of 3 × 3 matrix:



a b c
e f d f d e
|A| = d e f = a − b + c


h i g i g h
g h i

Notion In eq. each determinant of a 2 × 2 matrix is called a minor of the matrix A.

Remark 1: A similar procedure can be applied if the size of the matrix is larger.
Remark 2: A general formula for the determinant of a square matrix can be found elsewhere
(cf. wikipedia)

Definition of a first minor


If A is a square matrix, then the minor of the entry in the row i and column j (also called the

15
(i, j) minor, or a first minor) is the determinant of the submatrix formed by deleting the i-th
row and j-th column. This number is often denoted Mi,j .

General definition of a minor of a k th order


Let A be an m × n matrix and k an integer with 0 < k ≤ m, and k ≤ n. A k × k minor of A
(also called minor determinant of order k of A) is the determinant of a k × k matrix obtained
from A by deleting m − k rows and n − k columns.

Definition: The rank of A is the largest order of any non-zero minor in A.

16
Linear systems
The solutions of linear systems is likely the single largest application of matrix theory.
Definition of a linear system of equations
Let the coefficients aij , 1 ≤ i ≤ m and 1 ≤ j ≤ n and the data bl , 1 ≤ l ≤ m are known. We
define the linear system for the n unknowns x1 , ·, xn to be:

a11 x1 + a12 x2 + · · · + a1n xn = b1


a21 x1 + a22 x2 + · · · + a2n xn = b2
···
am1 x1 + am2 x2 + · · · + amn xn = bm

This system can be written in a form of matrix equation

Ax = b

with
     
a11 a12 ··· a1n x1 b1
 a21 a22 ··· a2n   x2   b2 
     
A=
 .. .. .. ,
..  x=
 .. ,
 b=
 .. 

 . . . .  .  . 
am1 am2 · · · amn xn bm
Definition of linearly dependent equations. An equation is said to be linearly dependent
with others in the system if it can be obtained from them with operation of multiplication by
a scalar and summation.

Solution of a linear system and its geometrical interpretation


A solution of a linear system is an assignment of values to the variables x1 , x2 , ..., xn such that
each of the equations is satisfied. The set of all possible solutions is called the solution set.

Geometric interpretation for solution set


Consider a system of equations for two variables x and y. For a system involving two variables
(x and y), each linear equation determines a line on the xy-plane (fig. 4a). So, if only one
equation is given for two variables, the solution set is infinite, it presents a line.
Consequently, if the second equation exists in the system, it may correspond to a line which
intersects the first one or which is parallel to the first one (fig. 4b). In case of intersection, the
point is a solution for the system of two equations. If the lines are parallel, no solution exists
for given two equations.
Let an interaction has was found but a third equation exist in the system which corresponds
to a third line. This line can intersect the two given before either in the same point or not. In
17
Figure 4: Graphical interpretation of solution of a linear system for two variables with two
equations: left and right figures underestimated systems, center: well defined (consistent)
system

the first case the solution set does not change, it corresponds to the found point while in the
second the solution set is empty, see Fig.5.

Figure 5: Graphical interpretation of solution of a linear system for two variables with three
equations. At the left overestimated systems, but any system of two equations gives the same
solution; at the right: overestimated inconsistent system

Consequently, we found three situations:


1. Number of equations less then number of unknowns. Then, usually there is infinitely many
solutions. The system is called undetermined.
2. Number of equations equal to the number of unknowns. If among the equations no linearly
dependent equations exist, the system has one and only one solution.
3. Number of equations larger then the number of unknowns, suppose that the system does
not contain linearly dependent equations. Then a reduced system with number of equations
equal to the number of unknowns can be solved (any equations can be chosen), the obtained
solution set should be put into remained equations to verify that the latter are satisfied.

A necessary and sufficient condition for the existence of a solution set for the system
of linear equation is the following:
A system of linear equations has one and only one solution set if it consists of linearly indepen-
dent equation, number of which is equal to the number of unknowns.

Equivalent formulation of a necessary and sufficient condition for the existence of


a solution set for the system of linear equation:
A system of linear equations has one and only one solution set if the rank of the matrix A is

18
equal to the number of independent variables.

Example: consider a system with a matrix A of a size n × n, with n unknowns and the vector
of free members b is not full of zeros

An x = b

It will have one and only one solutions set if det |A| =
6 0.

Finding a solutions of systems of inhomogeneous linear equations


Gauss elimination

In theory, solution of a system can be obtained using Gauss elimination which is based on three
operations:
(GE1) Interchange two equations
(GE2) Multiply any equation by a nonzero constant
(GE3) Add a multiple of one equation to another

It can be proved that these operations do not change the solution set of the systems. The idea
is to get from an initial matrix A an upper-triangle matrix (fig.) which allows easily find the
unknowns.
    
a11 a12 a12 · · · a1n x1 b1
    
 0 a22 a23 · · · a2n  x2   b2 
    
A=
0 0 a 33 · · · a 3n  x3  =  b3 
   
 . .. .. . . .. . .
 .. . . . . . .
  .   . 
0 0 0 · · · amn xn bn

Cramer’s rule

Alternatively, Cramer’s rule can be used which says that if a system of equations is given in a
matrix form
Ax = b

and if det(A) 6= 0, then the system has a unique solution, whose individual values for the
unknowns are given by:
det(Ai )
xi = i = 1, . . . , n
det(A)
where Ai is the matrix formed by replacing the i-th column of the matrix A by the column
vector b.

19
Solutions of systems of homogeneous linear equations

A system of linear equations where the right part (free members) is equal to zero (cf.20) is
called a system of homogeneous linear equations.

Ax = 0 (20)

From consideration above we know that if det(A) 6= 0 in 20 then the given system has one
and only one solution set which is (obviuosly) x ≡ 0. The solution set which contains only
zeros is called ”trivial solution”. Consequently, to a necessarily condition that the system 20
would have non-trivial solution is det(A) = 0 and in this case there are infinitely many solution
sets which have the following properties:
· If u and v are two vectors representing solutions to a homogeneous system, then the vector
sum u + v is also a solution to the system.
· If u is a vector representing a solution to a homogeneous system, and r is any scalar, then ru
is also a solution to the system.

Remarques on numerical solution of systems of linear equations


Although the methods and equations given above allows one to solve the system of linear equa-
tions readily, in numerical methods where large matrices are generally obtained, they are not
applicable. In particular, calculation of determinant take too long time. Gauss technique is
used but it is improved using so-called relaxation parameters.
In short, generally iterative schemes are used for solution of the linear system of equations
which allows parallelization of the algorithms. The list of methods is given below
Gauss-Zeidel

20
Exercises with matrices and linear systems
Matrix transposition
Transpose the following matrices:
 
x1
 x2 
 
 
8x 5y 12xy  
 x3 
A = 28z 43x 87y  X=  (21)
   
 x4 
4 11 5xy .
.
.
xn
 
a11 a12 a13 0 0 0
A = a21 a22 a23 0 0 0 (22)
 

a31 a32 a33 0 0 0

System of linear equation


Verify if the system of linear equation Ax = B with matrices A and B given below has unique
solution and if yes, solve it using Gauss elimination
 
2 2 −1
A = 0 1 3  (23)
 

1 2 −1
 
−61
B= 2  (24)
 

−13

Block Matrices
Sometimes it is convenient work with partitioned or blocked matrices, that are matrices whose
entries are themselves matrices. Let us try to see an example. Let I2 is an identity 2 × 2
matrix. Let the matrix A is expressed as:
" #
aI2 cI2
A= (25)
dI2 eI2

Write down the matrix A in its full form.

21
Derivatives
Limit of a function
To start with, first of all we should refresh the definition of the limit of a function which is the
following.
Definition of the limit of a function Suppose f : R → R is defined on the real line and
p, L ∈ R. It is said the limit of f , as x approaches p, is L and written

lim f (x) = L (26)


x→p

if the following property holds: For every real ε > 0, there exists a real δ > 0 such that for all
real x condition 0 < |x − p| < δ implies |f (x) − L| < ε. The value of the limit does not depend
on the value of f (p), nor even that p be in the domain of f .

Example
The function
sin x
f (x) = (27)
x
is not defined at x = 0 but the limit there exists and is equal to 1.

Example
Limits can also have infinite values, example is the function f (x) = 1/|x| where at x = 0 the
function is not defined and limits from the left and from the right (i.e. for x > 0 and x < 0)
are both equal to infinity.

Calculation of a limit of a function Among one of properties of the limit the one is of
special interest for us and which actually serves for the definition of a continuous function: The
function f is continuous at p if and only if the limit of f (x) as x approaches p exists and is
equal to f (p).

Consequently, for continuous functions the value of a limit at a certain point is simply a value
of the function at this point.
Definition of a derivative
The derivative of f (x) with respect to x (in any point) is the function f 0 (x) which is defined
as a limit
f (x + h) − f (x)
f 0 (x) = lim (28)
h→0 h
If this limit exist for any point where the function is defined, we say that the function f (x)
is differentiable everywhere. When we say that the limit exist, we assume that both, left and
right limits exist and they are equal.

22
Example of a derivative calculation via definition 1: Consider example f (x) = x2 . Let’s
check if the function is differentiable everywhere. Apply the definition

f (x + h) − f (x) (x + h)2 − x2
f 0 (x) = lim = lim =
h→0 h h→0 h
2xh + h2
= lim = 2x
h→0 h
Another example of a derivative calculation via definition: Consider a function

x2 , if x ≤ 1
f (x) = (29)
x, if x 1

According to the previous example, for all points x < 1 the derivative exists and is equal to 2x.
For all points x > 1 according to definition, the derivative is

f (x + h) − f (x) (x + h) − x h
f 0 (x) = lim = lim = lim = 1
h→0 h h→0 h h→0 h

Consequently, in the point x = 1 the limit from the left side (x → 1 − 0) exists and equal to
2, but the limit from the right side (x → 1 + 0) is equal to 1. The limits are not equal, the
function has no derivative in the point x = 1.

Definition of a differential Let the function y = f (x) is defined in a neighborhood of a point


x and dx is a variation (increase) of the independent variable x. It is said that the function
y = f (x) has a (first) differential in the point x if its variation at this point can be presented as

δy ≡ f (x + dx) − f (x) = Adx + o(dx) (30)

where A does not depend on dx and o(x) is so-called ”little o”. In this case the linear part
dy = Adx is called ”differential”.
Note that the function y = f (x) has a (first) differential at the point x if and only if it has a
(first) derivative at this point defined as dy/dx = f 0 (x), so that δy = f 0 (x)dx + o(dx).
Using definition given by eq.28, it is possible to define some key derivatives, list of them is
provided at the end of this chapter.

Rules to calculate derivatives


To calculated derivatives for other functions, the following rules, which follow from the
definition of the derivative and calculations of limit can be used :

Summation rule
d [f (x) + g(x)] df dg
[f (x) + g(x)]0 = = f 0 (x) + g(x)0 = + (31)
dx dx dx

23
Product rule
d [f (x) · g(x)]
[f (x) · g(x)]0 = = f 0 (x)g(x) + f (x)g(x)0 (32)
dx
As a consequence, a rule for the case of a product with a constant

Chain rule
d[f (g(x))]
(f [g(x)]))0 = = f 0 [g(x)] · g 0 (x) (33)
dx
As a consequence, a quotient rule (demonstrated in slides)

Interpretation of derivatives
Slope of the tangent line

One of the major interpretations of the derivative f (x) at a point x0 is the slope of the tangent
line to f (x) at x0 (cf fig. 6). Furthermore, the slope of the line perpendicular to the tangent
(or the line which is ”normal” to the curve) at the point x0 can be obtain.

Figure 6: Illustration of a tangent line and a perpendicular line to f (x) = x2 at the point x = 1

Use of derivatives to study the properties of the function

Example 1: Sketch the graph of a derivative for the function given at fig.7. Steps to do:
1. Define the ”critical points” that are points where the line tangential to the curve is horizon-
tal. At these points the derivative is equal to zero.
2. Define if the function increases or decreases between the critical points found during the
first step: this will provide the sign of the derivatives between the critical points

Example 2: study the property of the function y(x) = x(ln x)2 and sketch this function
(demonstrated in slides)
24
Figure 7: Sketch a graph for derivative of the given function

Figure 8: A graph of derivative of the function given in fig. 7. What about the second
derivative?

Implicit derivation
A functions can be given in a form not resolved for y(x), i.e. f (x, y) = g(x, y) examples are:

yx = 1
x2 + y 2 = 9
x3 y 5 + 3x = 8y 3 + 1
··· (34)

How to find a derivative y 0 (x)?


Use a chain rule and then solve for y 0 (x) (if possible!) (examples are given in the slides)

25
e2x+3y = x2 − ln(xy 3 )
0 2x+3y (y 3 + 3xy 2 y 0 )
(2 + 3y )e = 2x −
xy 3
1 3y 0 )
2e2x+3y + 3y 0 e2x+3y = 2x − −
x y
1 1
3y 0 (e2x+3y − ) = 2x − 2e2x+3y −
y x
(35)

26
Table of Key Derivatives
Table of derivatives

dxα
= αxα−1
dx
dex
= ex
dx
dax
= ax ln a
dx
dln(x) 1
=
dx x
dC
=0
dx
(36)

Derivatives of trigonometric functions

d sin x
= cos x
dx
d cos x
= − sin x
dx
d tan x 1
= = (secx)2
dx (cos x)2
d cot x 1
=− = (−cscx)2
dx (sin x)2
(37)

Derivatives of the inverse trigonometric functions

Let

x = sin y y = arcsin x

then
d arcsin x 1
=p
dx (1 − x2 )
d arccos x 1
= −p
dx (1 − x2 )
d arctan x 1
=
dx 1 + x2
(38)

27
Derivatives of functions of multiple variables
Similar to the definition of a limit of a function of one variable x, we can introduce definition
of a limit for the function of multiple variables. A simple formulation is as following

Definition of a limit for a function of two variables. Let f (x, y) : R → R. L is said to


be a limit of f in the point (x, y) if f (x, y) approaches L when (x, y) approaches (a, b) along
any direction.
Definition The function f (x, y) is continuous in a point (x0 , y0 ) if lim(x,y)→(x0 ,y0 ) = f (x0 , y0 ).
Example
x2 y 2 9
lim 4 4
= (39)
(x,y)→(1,3) x + 3y 244
Yet, the limit at the point (0, 0) does not exist. It is proved by considerations of different paths
of approaching to (0, 0): along x, with y = 0, along y with x = 0 and along the line x = y:
x2 y 2 x2 02
lim = lim =0
(x,y)→(0,0) x4 + 3y 4 (x,0)→(0,0) x4 + 3 · 04

x2 y 2 02 y 2
lim = lim =0
(x,y)→(0,0) x4 + 3y 4 (0,y)→(0,0) 04 + 3y 4

x2 y 2 x2 x2 1
lim 4 4
= lim 4 4
=
(x,y)→(0,0) x + 3y (x,x)→(0,0) x + 3y 4

Partial derivatives
Definition of partial derivatives
Two first order partial derivatives can be defined for the function of two variables as following:
∂f (x, y) f (x + h, y) − f (x, y)
fx0 (x, y) = = lim (40)
∂x h→0 h
f (x, y + h) − f (x, y)
fy0 (x, y) = lim (41)
h→0 h
Example

f (x, y) = x4 − 10x2 y 2 + 6 y − 10
fx0 (x, y) = 4x3 − 20xy 2
1
fy0 (x, y) = −20x2 y + 3 √
y
Definition of the differential
Definition of the differential of a function of multiple variable (here we have only two) is similar
to the case of one variable. Let the function z = f (x, y) is defined in a neighborhood of a point
(x, y) and dx, dy are variations (increase) of the independent variables x and y. It is said that
the function z = f (x, y) has a (first) differential in the point (x, y) if its variation at this point
can be presented as

δz ≡ f (x + dx, y + dy) − f (x, y) = Ax dx + Ay dy + o(ρ) (42)


28
p
where Ax , Ay do not depend on dx, dy and ρ = (x2 + y 2 ) . In this case the linear part
dz = Ax dx + Ay dx is called ”differential”.
Note that the function z = f (x, y) has a (first) differential at the point (x, y) if and only if it has
continuous (first) derivatives at this point defined as ∂z/dx = fx0 (x, y) and ∂z/dx = fy0 (x, y),
0
so that δz = fx,y (x)dx + fy0 (x, y)dx + o(ρ).
Application of partial derivatives
·Generally, fx0 (x, y) represents the rate of change of the function f (x, y) as we change x and
hold y fixed.
· fy0 (x, y) represents the rate of change of f (x, y) as we change y and hold x fixed.

Chain rule for the function of two variables. Jacobian matrix


Case z=f(x,y), x=g(t), y=h(t), dz/dt?

The function f (x, y) can be represented as z = f (g(t), h(t)). Then the ”simple” chain rule is
applicable:
dz d[f (x, y)] ∂f (x, y) dx ∂f (x, y) dy
= = +
dt dt ∂x dt ∂y dt
Example
z = x2 y 3 + y cos x, x = ln(t2 ), y = sin(4t). Calculate dz/dt

dz ∂z dx ∂z dy 2
= + = (2xy 3 − y sin x) + (3x2 y 2 + cos x)(4 cos t)
dt ∂x dt ∂y dt t
Now, we can express everything in t or leave with x, y, accounting for t = ex/2 or t = 1/4 arcsin(t)
Example
z = x ln(xy) + y 3 , y = cos(x2 + 1). Calculate dz/dx.
Note that dy = −2x sin(x2 + 1)
   
dz ∂z dx ∂z dy y x
= + = ln(xy) + x + x + 3y (−2x sin(x2 + 1)) =
2
dx ∂x dx ∂y dx xy xy
 
x
(ln(xy) + 1) + + 3y 2 (−2x sin(x2 + 1))
y

Here y can be replaced with x.

Case z=f(x,y), x=g(s,t), y=h(s,t); ∂ z/∂ t and ∂ z/∂ s?

Here we have simple extension of the chain rule: if z = f (x, y), x = g(s, t), y = h(s, t),
then, obviously, z is a function of s and t and corresponding partial derivatives are:
∂z ∂z ∂x ∂z ∂y
= +
∂s ∂x ∂s ∂y ∂s
∂z ∂z ∂x ∂z ∂y
= +
∂t ∂x ∂t ∂y ∂t

29
Example
For the function z = e2r sin(3θ) compute partial derivatives ∂z/∂r and ∂z/∂θ.
√ ∂z ∂z
Now, let r = st − t2 , θ = s2 + t2 . Compute and :
∂s ∂t
∂z ∂z ∂r ∂z ∂θ s
= + = 2e2r sin(3θ) · t + 3e2r cos(3θ) √
∂s ∂r ∂s ∂θ ∂s s2 + t2
∂z ∂z ∂r ∂z ∂θ t
= + = 2e2r sin(3θ)(s − 2t) + 3e2r cos(3θ) √ (43)
∂t ∂r ∂t ∂θ ∂t s2 + t2
here either r, θ should be replaced with s, t or vice versa
Example
Let us consider a function z = 5r2 and let us know that x = r cos θ and y = r sin θ. Find ∂z/∂x
and ∂z/∂y.
Solution Using chain rule, we can write:
∂z ∂z ∂r ∂z ∂θ ∂z ∂z ∂r ∂z ∂θ
= + = +
∂x ∂r ∂x ∂θ ∂x ∂y ∂r ∂y ∂θ ∂y
(44)

Let us write and for the differentials dx and dy:


∂x ∂x
dx = dr + dθ (45)
∂r ∂θ
∂y ∂y
dy = dr + dθ (46)
∂r ∂θ
This can be written as well in the matrix form:
" # " #
dx dr
=J (47)
dy dθ

where the matrix  


∂x ∂x " #
 ∂r ∂θ  cos θ (−r sin θ)
Jx,y (r, θ) =  ∂y ∂y  = (48)
sin θ r cos θ
∂r ∂θ
is so-called Jacobian matrix, which is for this particular case corresponds to the transition from
the Cartesian to polar coordinate system. Note that the system of equations 45-46 can be
solved for dr and dθ using Cramer rule giving:

dx −r sin θ cos θ dx


dy r cos θ x y sin θ dy −y x
dr = = dx + dy dθ = = 2 dx + 2 dy
|Jx,y (r, θ)| r r |Jx,y (r, θ)| r r
Then the Jacobian matrix for the transition from polar to Cartesian coordinates is
 x y
r
Jr,θ (x, y) =  −y r
x
r2 r2
Note that
30
Jx,y (r, θ)Jr,θ (x, y) = I

The determinant of the Jacobian matrix det(J) = |J| is called Jacobian. There are different
notations for Jacobians, for example:
g g

(g, h) t s
= h h
t, s
t s
Finally, let us put everything together. We note that if we put partial derivatives in a form of
a vector (∂z/∂x, ∂z/∂y), then we can write:
−1
(∂z/∂x, ∂z/∂y) = (∂z/∂r, ∂z/∂θ)Jr,θ (x, y) = (∂z/∂r, ∂z/∂θ)Jx,y (r, θ)

where, as we discussed earlier, a matrix in the power −1 means the ”inverse matrix”.

Chain rule for multiple variables

Suppose that z is a function of n variables, x1 , x2 , x3 , · · · , xn , and that each of these variables


are in turn functions of m variables, t1 , t2 , t3 , · · · , tn . Then for any variable ti we have the
following:
n
∂z ∂z ∂x1 ∂z ∂x2 ∂z ∂xn X ∂z ∂xk
= + + ··· + =
∂ti ∂x1 ∂ti ∂x2 ∂ti ∂xn ∂ti k=1
∂xk ∂ti

Directional derivative
The rate of change of f (x, y) in the direction of the unit vector ~u = (u1 , u2 ) is called the
directional derivative and is denoted by D~u f (x, y). The definition of the directional derivative
is:
f (x + u1 h, y + u2 h) − f (x, y)
D~u f (x, y) = lim (49)
h→0 h
A more ”practical” formula to compute a directional derivative in case of two variables is:

D~u f (x, y) = fx0 (x, y)u1 + fy0 (x, y)u2 (50)

and in case of three variables if the directional vector is given as ~u = (u1 , u2 ))

D~u f (x, y, z) = fx0 (x, y, z)u1 + fy0 (x, y, z)u2 + fz0 (x, y, z)u3 (51)

Example: Compute D~u f (2, 0) for f (x, t) = xexy +y and ~u is a vector in the direction θ = 2π/3.

Solution: Using Cartesian coordinate, we can write for the directional vector ~u = (−1/2, 3/2).
The directional derivative is

−1 xy xy 3 2 xy
D~u f (x, y, z) = (e + xye ) + (x e + 1)
2 2
To finish the calculation, the values x = 2 and y = 0 should be put into the previous formulae.
Note that the eq.50 and eq.51 can be written using the scalar product (” dot product”) of two
vectors:
D~u f (x, y, z) = (fx0 (x, y, z), fy0 (x, y, z), fz0 (x, y, z)) · (u1 , u2 , u3 )
31
The second vector is a vector giving a direction along which the derivative has to be taken
whereas the first vector has special notation which is called gradient of the function f (x, y, z):

~ = (f 0 (x, y, z), f 0 (x, y, z), f 0 (x, y, z))


∇f x y z

where ∇~ is a special notation for the ”set of partial derivatives” which in Cartesian coordinates
is written as
~ = e~x ∂ + e~y ∂ + e~z ∂
∇ (52)
∂x ∂y ∂z
Exercise: Let a function z = f (x, y) is given and x = x(t, s) and y = y(t, s). Then the gradient
~ s,t f (x, y) can be calculated using the gradient of
of the function f over the coordinates (s, t) ∇
the given function over the coordinates (x, y) ∇~ x,y f (x, y) and a Jacobian matrix.

32
Taylor’s theorem, Taylor’s polynomial, Taylor and MacLau-
rin Series
Let us consider a function f (x) = sin x. According to the presentation with a circle, (see fig.9a),

Figure 9: Illustration to the definition of the sin α, cos α

y y
sin α = =p (53)
h x2 + y 2

Using (53), it is easy to obtain a value of the sinus/cosinus for α = 0. What if we wish to
obtain a value of sinus/cosinus for α = 30 = π/60?

Another example: f (x) = ex − x3 for x not very far from zero.

Taylor’s theorem in one real variable


Let k ≥ 1 be an integer and let the function f : R → R be k times differentiable at the point
a ∈ R. Then there exists a function hk : R → R such that

0 f 00 (a) 2 f (k) (a)


f (x) = f (a) + f (a)(x − a) + (x − a) + · · · + (x − a)k + hk (x)(x − a)k (54)
2! k!
and limx→a hk (x) = 0. Here hk (x)(x − a)k is a remainder in Peano form
The polynomial which appears in Taylor’s theorem is the k-th order Taylor polynomial

0 f 00 (a) 2 f (k) (a)


Pk (x) = f (a) + f (a)(x − a) + (x − a) + · · · + (x − a)k (55)
2! k!
and
Rk (x) = f (x) − Pk (x) (56)
is a remainder term which can be given in different forms. One of the presentation is Lagrange
form:
f (k+1) (ξL )
Rk (x) = (x − a)k+1 (57)
(k + 1)!
where ξL is a point between x and a. Yet, the form of the remainder is not important but it is
important to be able to estimate its value! Let us do it.
Suppose that we wish to approximate a function f (x) on the interval I and f (x) is k + 1 -times
33
continuously differentiable. The point a belongs to the interval I. Then, if |f k+1 (x)| ≤ M at
the interval I = (a − r, a + r) with some r > 0, we can estimate

(x − a)k+1 rk+1
|Rk (x)| ≤ M ≤M (58)
(k + 1)! (k + 1)!
Example Let us approximate with Taylor’s polynomial the function f (x) = sin x at a = 0
with the error of approximation less than 10−8 up to order k = 7. For which interval this will
be valid? Sinus approximation with Taylor’s polynomial up to 7 order at 0 is:
1 3 1 5 1 5
sin x = x − x + x − x + R7 (x)
3! 5! 7!
Remainder is estimated as
f (7+1) (ξL ) f (8) (ξL ) 8
R7 (x) = (x − 0)7+1 = (x) < 10−8
(7 + 1)! (8)!

Obviously, the maximal possible value of sin(8) (ξL ) is 1 whatever is ξL , so the condition is:
1
(r)8 < 10−8
(8)!

that gives r < (8· 10−8)1/8 , i.e. r < 0.3672029. We can check: the value of sin(0.3672029) =
0.3709204694 and the value P7 (x) = 0.3709204690. As we can see, we underestimated the
interval at which the difference between the Taylor’s polinomial P7 and the function itself still
satisfies imposed condition because the maxiaml value for f (ξL ) was taken too high. The
approximation with P7 is valid up to x ≈ 0.5
Example (from https://en.wikipedia.org/wiki/Taylor%27s_theorem) Suppose that
we wish to approximate the function f (x) = ex on the interval [1, 1] while ensuring that the
error in the approximation is no more than 105. In this example we pretend that we only know
d x
the following properties of the exponential function: (∗) e0 = 1, dx
e = ex , ex >
0, x∈R.
We know that f (k) (x) = ex for all k, and in particular, f (k) (0) = 1. Hence the k −th order
Taylor polynomial of f at 0 and its remainder term in the Lagrange form are given by
x2 xk eξ
Pk (x) = 1 + x + + ··· + , Rk (x) = xk+1 ,
2! k! (k + 1)!
where ξ is some number between 0 and x. Since ex is increasing by (*), we can simply use
ex ≤ 1 for x ∈ [1, 0] to estimate the remainder on the subinterval [1, 0]. To obtain an upper
bound for the remainder on [0, 1], we use the property e < ex for 0 < < x to estimate

eξ 2 ex
ex = 1 + x + x < 1 + x + x2 , 0<x≤1
2 2
using the second order Taylor expansion. Then we solve for ex to deduce that
1+x 1+x
ex ≤ x2
=2 ≤ 4, 0≤x≤1
1− 2 2 − x2
34
simply by maximizing the numerator and minimizing the denominator. Combining these esti-
mates for ex we see that
4|x|k+1 4
|Rk (x)| ≤ ≤ , −1 ≤ x ≤ 1,
(k + 1)! (k + 1)!
so the required precision is certainly reached, when
4
< 10−5 ⇔ 4 · 105 < (k + 1)! ⇔ k ≥ 9.
(k + 1)!
As a conclusion, Taylor’s theorem leads to the approximation
x2 x9
ex = 1 + x + + ... + + R9 (x), |R9 (x)| < 10−5 , −1 ≤ x ≤ 1.
2! 9!
For instance, this approximation provides a decimal expression e ≈ 2.71828, correct up to five
decimal places.

Taylor’s and MacLauren series


A Taylor series is a representation of a function as an infinite sum of terms that are calculated
from the values of the function’s derivatives at a single point. If the Taylor series is centered
at zero, then that series is also called a Maclaurin series.

Let I ⊂ R be an open interval. By definition, a function f : I → R is real analytic if it is


locally defined by a convergent power series. This means that for every a ∈ I there exists some
r > 0 and a sequence of coefficients ck ∈ R such that (ar, a + r) ⊂ I and

X
f (x) = ck (x − a)k = c0 + c1 (x − a) + c2 (x − a)2 + · · · , |x − a| < r. (59)
k=0

In general, the radius of convergence of a power series can be computed from the Cauchy-
Hadamard formula (see short discussion below).

The eq.59 represent a Taylor’s series for the function f (x). If the point a = 0 then this series
also has a special title: M acLaurenseries.

One can think about Taylor polynomials as truncation of Taylor’s series.


Consequently, we cans present sin x, cos x, ex as infinite sums and for these functions the radius
of convergence is infinity (the proof is out of the scope of this course)

sin x =
cos x =
ex =

Yet, there is a question: is it possible that the sum of infinite number of terms is still finite?
Yes. To check if a series converges one can use D’Alambert ratio test.
35
D’Alambert ratio test

Consider a series {an }. It converges if there exists a limit:



an+1
ρ = lim =q (60)
n→∞ an

and q < 1.

Hence for the sin x we have:


x2k+3 (2k + 1)! x2
lim = lim → 0 ∀x (61)
k→∞ (2k + 3)! x2k+1 k→∞ (2k + 3)(2k + 2)

36
Indefinite integral or anti-derivative or primitive
Given a function f (x), an anti-derivative (or a primitive function) of f (x) is any function F (x)
such that
F 0 (x) = f (x) (62)

Taking into account that the derivative of a constant value is zero, it is evident that F (x) + C
where C is any constant is also anti-derivative of f (x). If F (x) is any anti-derivative of f (x)
then the most general anti-derivative of f (x) is called an indefinite integral and is denoted
as: Z
F (x) = f (x)dx + c (63)

Using knowledge from the section of derivative, we can guess anti-derivatives (or primitive)
functions for some basics:
Z
Cdx = Cx + C1
xα+1
Z
xα dx = + C, if α 6= −1
α+1
Z Z
−1 1
x dx = = lnx + C
x
Z
cos xdx = sin x + C
···
(64)

Properties of the indefinite integrals


Z Z
k · f (x)dx = k f (x)dx (65)
Z Z Z
[f (x) ± g(x)] dx = f (x)dx ± g(x)dx (66)

The following terminology is used:


dx is a differential
f (x) is the integrand (a function to be integrated) Z
F (x) is anti-derivative or primitive or indefinite integral of the function f (x) is integration
(taking the integral), the operation inverse to derivation (taking of the derivative).

Generally, integration of a function is made using the knowledge of the ”basic” integrals
(or, similar: basic derivatives) and some rules described below.

Substitution rule
Z Z
0
f [g(x)] g (x)dx = f (u)du with u = g(x) (67)

37
Indeed: du/dx = d[g(x)]/dx so du = g 0 (x)dx
Example: Z Z
1 1 1
sin 2xdx = sin udu|u=2x = − cos u = − cos 2x + c
2 2 2
Indeed: u = 2x then du = 2dx and dx = du/2
Example:
Z Z
2 3 4 1 1 1
x (3 − 10x ) dx = − (3 − 10x3 )4 d(3 − 10x3 ) = − · (3 − 10x3 )5 + c
30 30 5
Here intermediate steps with the substitution u = (3 − 10x3 ) and du = −30x2 dx were not done
explicitly.
Example (easy): Z
(sin 2x)3 cos 2xdx

Example: Z
tan xdx

use u = cos x, du = − sin x, the integrals becomes:


Z Z
du
tan xdx = − = − ln u + C = − ln(cos x) + C
u
Z √
Example: 2x x2 + 1dx Using u = x2 + 1, du = 2x
Z √ Z
√ 2 2
2x x2 + 1dx = udu = u3/2 + C = (x2 + 1)3/2 + C
3 3

Integration by parts
Note that

[f (x)g(x)]0 = f 0 (x)g(x) + f (x)g 0 (x)


f (x)g 0 (x) = [f (x)g(x)]0 − f 0 (x)g(x)
Z Z Z
[f (x)g (x)] dx = [f (x)g(x)] dx − f 0 (x)g(x)dx
0 0

Z Z
[f (x)g (x)] dx = f (x)g(x) − f 0 (x)g(x)dx
0

Z Z
udv = uv − vdu

Example:
Z Z Z
x cos xdx = xd(sin x) = x · sin x − sin xdx = x · sin x + cos x + C (68)

Example:
Z Z Z Z
x e dx = x de = x e − e d(x ) = x e − 2 ex xdx =
2 x 2 x 2 x x 2 2 x

Z Z
x e − 2 xd(e ) = x e − 2[xe − ex dx] = x2 ex − 2xex + 2ex + C
2 x x 2 x x
(69)

38
Another way to take the proposed integral is the following:
Suppose that the solution is of the form I = (a0 + a1 x + a2 x2 )ex , so, it’s derivative must be
equal to x2 ex . We get:

dI/dx = (a1 + a0 )ex + (a1 + 2a2 )xex + a2 x2 ex

that gives us conditions: a2 = 1, a1 + 2a2 = 0, a1 + a0 = 0, i.e. a1 = −2, a0 = 2. That leads to


the solution: I = ex (x2 − 2x + 2) + C.

Recomendations for the integration by parts LIATE / DETAIL rules


Suppose that P (x) is a polynome of degree n ≥ 0 then for the integrals
Z
P (x) ln xdx
Z
P (x) arctan dx
Z
P (x) arccos xdx
Z
P (x) arcsin xdx
Z
the choice is dv = P (x)dx, so that v = P (x).
Example: Z
xn ln xdx, n 6= −1

Taking dv = xn dx, we know that v = xn+1 /(n + 1). Consequently, u = ln x. Then obtain:

xn+1
Z Z Z Z
n 1 1
x ln xdx = udv = uv − vdu = ln x − xn+1 dx =
n+1 n+1 x
n+1 Z
x 1
ln x − 2
xn+1 + C (70)
n + 1 (n + 1)
for the integrals
Z
P (x) sin xdx
Z
P (x) cos dx
Z
P (x)ex dx
Z
P (x)ax dx
(71)

the choice is u = P (x), so that du = P 0 (x).

39
Z
A nice example of the integration xn sin(x)dx can be found in the internet:
https://math.stackexchange.com/questions/231100/computing-the-indefinite-integral-int-xn-sin-x-dx

This recommendation is known as LIATE rule of thumb rule proposed by Herbert Kasube
advises that functions comes first in the following list should be chosen as u: L - Logarithmic
Functions: ln(x), logb (x), etc.
I - Inverse trigonometric functions: arctan(x), arcsec(x), etc.
A - Algebraic functions: x2 , 3x50 etc
T - Trigonometric functions: sin(x), tan(x) etc
E - Exponential functions: ex , 19x , etc
The function which is to be dv is whichever comes last in the list: functions lower on the list
have easier antiderivatives than the functions above them. The rule is sometimes written as
”DETAIL” where D stands for dv.

Recommended substitutions for indefinite integral


Example: Z
sinm x cosα xdxfor any αand if m=2k+1 and

Use the substitution cos x = t with dt = − sin xdx, i.e. dx = −dt sin−1 x and identity
sin2 x + cos2 x = 1, i.e. sin2 x = 1 − cos2 x = 1 − t2 then
Z Z Z
m α 2 (m−1)/2 α
sin x cos xdx = (1 − t ) t dt = (1 − t2 )(m−1)/2 tα dt =
Z
(1 − t2 )k tα dt (72)

Integration of R(sin x, cos x)

Recommended so-called ”universal substitution” is u = tan(x/2) which gives

2u 1 − u2 2du
sin x = , cos x = , dx =
1 + u2 1 + u2 1 + u2
Example

2du 1 + u2
Z Z Z
dx du
= = = ln|u| + C = ln|tan(x/2)| + C
sinx 1 + u2 2u u
Yet, sometimes this substitution can lead to complicated functions. Other substitutions are
recommended: sin x = t, cos x = t, tan x = t.

40
Integration of R(shx, chx)

Reminder:
ex − e−x ex + e−x sinh x e2x − 1
shx = sinh x = , chx = cosh x = , tanh x = = 2x
2 2 cosh x e +1
Recommended substitution is u = th(x/2) which gives also

2u 1 + u2 2du
shx = , chx = , dx =
1 − u2 1 − u2 1 − u2
√ √ √
Integration of R(xand 1 − x2 , x2 − 1, x2 + 1)

Integration of these functions is reduced to the previous case with the substitution x = cos v
or x = shv or x = shv. Recommended substitution is u = th(x/2) which gives also

2u 1 + u2 2du
shx = , chx = , dx =
1 − u2 1 − u2 1 − u2

41
Definite integral
Let f (x) is a function defined and limited at a segment [a, b]. Let us divide this segment on n
intervals with points a = x0 < x1 < x2 < ... < Xn = b. For each of these intervals let us choose
an arvitrary point ξi : (xi ≤ ξi ≤ ξi+1 ) and make a sum
n
X
f (ξi )(xi − xi1 ) (73)
i=1

If a limit of a the sum given by eq.73 when the length of the largest interval goes to zero exists,
then f (x) is said to be integrable in Riemann’s sense at the segment [a, b]. The limit
n
X Z b
I = limmax(xi −xi−1 )→0 f (ξi )(xi − xi1 ) = f (x)dx (74)
i=1 a

is a definite integral of f (x) on the closed interval [a, b] in Riemann’s sense (or Riemann’s
integral).

Figure 10: Illustration to the definition of the definite integral. Note that the given division
for the intervals would not correspond to the one providing a good value of the integral or the
area bounded by the f (x), x-axis and vertical lines x = a and x = b

This definition means that for any positive ε we can find δ > 0 such that for any arbitrary
division of the segment [a, b] on the sub-intervals with the length less than δ, i.e.

max(xi − xi−1 ) < δ

and for any arbitrary choice of the points ξi the following inequality fulfils
n
X
| f (ξi )(xi − xi1 ) − I| < ε
i=1

According to the definition, the definite integral of the function f (x) on the segment [a, b]
is a signed area bounded by the graph of the function f (x), the x-axis and the vertical lines
corresponding to the limits of the integral, x = a and x = b (Fig.10). Terminology: a and b
are limits of integration

42
Definite integral: properties
Z b Z a
f (x)dx = − f (x)dx (75)
a b
Z a
f (x)dx = 0 (76)
a
Z b Z b
c · f (x)dx = c · f (x)dx (77)
a a
Z b Z b Z b
[f (x) ± g(x)]dx = f (x)dx ± g(x)dx (78)
a a a
Z c Z c Z b
f (x)dx = f (x)dx + f (x)dx (79)
a a c
Great, but how to calculate a definite integral? Is it related to the indefinite integral that
we studied just before?

Fundamental theorem of calculus


From wiki: ”There are two parts to the theorem. Loosely put, the first part deals with the deriva-
tive of an antiderivative, while the second part deals with the relationship between antiderivatives
and definite integrals”.

Fundamental theorem of calculus: part I

If f (x) is continuous on [a, b] then,


Z x
F (x) = f (t)dt
a

is continuous on [a, b] and it is differentiable on (a, b) and

F 0 (x) = f (x)

Alternative notation for the last equation is


Z x
d
f (t)dt = f (x)
dx a
Example Differentiate the following:
Z t
g(x) = e2t cos(t3 + 5)dt
−4

In this example a dirrect application of the first part of the theorem is needed.
Example Differentiate the following:
Z 1 4
t +1
g(x) = 2
dt
x2 t + 1

43
Here the limit the integral is not given in a good form. Let us: (a) change the order of the
limits, (b) then substitute a ”good” variable use a chain rule, (c) put the variable x back into
formula
Step a:
Z 1 4 Z x2 4
dg(x) d t +1 d t +1
= 2
dt = − dt
dx dx x2 t + 1 dx 1 t2 + 1
Step b:

u = x2
dg(u) dg(u) du
=
u  dx du dx
t4 + 1 u4 + 1 du
 Z
dg(x) du d du
=− dt =− 2
du dx du 1 t2 + 1 dx u + 1 dx
Step c: account for that du/dx = 2x and put back x:
u4 + 1 du x8 + 1
− = −2x
u2 + 1 dx x4 + 1

Fundamental theorem of calculus: part II

Suppose f (x) is a continuous function on [a, b] and also suppose that F (x) is any anti-derivative
for f (x). Then,
Z b
F (x) = f (t)dt = F (x)|ba = F (b) − F (a)
a
Example:
8  8
x3 83 (−1)3
Z   
2
511
x + 30 = + 30x =
+ 30 · 8 − − 30 = + 270
−1 3 −1 3 3 3

Substitution rule for definite integrals


Example. Calculate the integral Z 0 √
2t2 1 − 4t3 dt
−2

Note that 2t = 1/3d(2t ) = 1/6d(4t ) = −1/6d(1−4t3 ). Let us make a substitution u = 1−4t3 .


2 3 3

Then du = 12t2 dt or dt = du/(12t2 ) and


Z 0 Z 0
2
√ du 1√ 1 2 3/2 t=0
2t u 2 = udu = u t=−2
−2 12t −2 6 63
At this point you have two options: (1) either to put back the variable t and to keep the limits,
or (2) to keep u as variable and reclaculate the limits appropriately:
1 2 3/2 t=0 1 1
· u t=−2 = u3/2 −33
6 3 9
or
1 2 3/2 t=0 1 0
· u t=−2 = (1 − 4t3 )3/2 −2
6 3 9

44
Multiple Integrals
Similar to the integral of one variable it is possible to introduce an integral for a function of two
variables, f (x, y). In case of 2 variables we will integrate over a region of R2 (two-dimensional
space). We will start out by assuming that the region in R2 is a rectangle which we will denote
as follows:

R = [a, b] × [c, d]

i.e. a ≤ x ≤ b and c ≤ y ≤ d (fig.11). Definition of the integral in R2 would be similar to that

Figure 11: Illustration to the integration over a rectangle


.

for one variable: we should subdivide the given rectangle into smaller non-overlapping regions,
for each of them choose points at the surface S whic will define the height of the corresponding
element and to calculated the volume. If, with decreasing the area of the constructed element
we have a limit for the volume, then the functionis integrable, etc. Fubinis Theorem
If f (x, y) is continuous on R = [a, b] × [c, d] then,
Z Z Z bZ d Z d Z b
f (x, y)dA = f (x, y)dydx = f (x, y)dxdy (80)
R a c c a

These integrals are called iterated integrals.


Example Z Z
1
2
dA, R = [0, 1] × [1, 2]
R (2x + 3y)

45
Let us first integrate over x:
2 1 2 1
−1 2
 Z  
−1
Z Z Z
−2 −1 1 1
(2x + 3y) dydx = (2x + 3y) dy = − dy =
1 0 1 2 0 2 1 2 + 3y 3y
 2
−1 1 1 −1
ln |2 + 3y| − ln |y| = (ln 8 − ln 2 − ln 5) (81)
2 3 3 1 6

Product under the integral sign and product of the integrals


If f (x, y) = g(x)h(y) and we are integrating over a rectangle R = [a, b] × [c, d], then:
Z Z Z Z Z b  Z d 
f (x, y)dA = g(x)h(y)dA = g(x)dx h(y)dy
R R a c

Example
Z Z
x(cos y)2 dA, R = [−2, 3] × [0, π/2]
R

Double integrals over general regions


Z Z
The region over which the integral f (x, y)dA has to be taken is not always a region. In
A
examples shown below the boundaries over one variable are defined as a function of another
one variable. For the first case the area of the integration and the integral are

Figure 12: Illustration to the definition of the definite integral. Note that the given division
for the intervals would not correspond to the one providing a good value of the integral or the
area bounded by the f (x), x-axis and vertical lines x = a and x = b

A = {(x, y) : a ≤ x ≤ b, g1 (x) ≤ y ≤ g2 (x)}


Z Z Z b Z g2 (x)
f (x, y)dA = f (x, y)dydx
A a g1 (x)

46
and for the second

A = {(x, y) : h1 (x) ≤ x ≤ h2 (x), c ≤ y ≤ d}


Z Z Z d Z h2 (x)
f (x, y)dA = f (x, y)dxdy
A c h1 (x)

Properties
Z Z of the double integral:
Z Z Z Z
1. (f (x, y) + g(x, y))dA = (f (x, y)dA + g(x, y))dA
Z ZA Z Z A A

2. cf (x, y)dA = c f (x, y)dA


A Z Z A Z Z Z Z
3. If A = A1 ∪ A2 , f (x, y)dA = f (x, y)dA1 + f (x, y)dA2
A A1 A2

Example:

Let A is the region bounded by x3 and x, one should take the integral 4xy − y 3 over the
region A. Solution:

1. The imits over y axis are given by equations y1 (x) = x3 and y2 (x) = x

2. The limits over the x-axis can be found via solution of the equality: x3 = x which gives
two roots: x1 = 0, x2 = 1.
3. The integral to take:
Z 1 Z √
x Z 1  √
4 x

y
(4xy − y 3 )dydx = 2xy 2 − dx =
0 x3 0 4 x3
Z 1 2  1
x12
 3
x8 x13

7x 7 7x 55
− 2x + dx = − + = (82)
0 4 4 12 4 52 0 156

Example:
Integrate 6x2 −40y over a triangle with vertices given by the coordinates (0, 3), (1, 1) and (5, 3).
Solution:
1. Define equations for lines bounded the region of integration as shown in the fig.13.
1.a. In terms of y as a function of x:

Figure 13: Illustration to the problem of integration of 6x2 − 40y over a triangle

47
for 0 ≤ x ≤ 1 y1 (x) = −2x + 3
for 1 ≤ x ≤ 5 y2 (x) = 1/2x + 1/2
for 0 ≤ x ≤ 5 y3 (x) = 3

1.b. In terms of x as a function of y

for 3 ≤ y ≤ 5: x1 (y) = −1/2y + 3/2 and x2 (y) = 2y − 1.

2. The integral can be taken using one of two ways:


Z 1 Z y=3 Z 5Z y=3
2
(6x − 40y)dydx + (6x2 − 40y)dydx (83)
0 y1 (x) 1 y2 (x)

or Z 3 Z x2 (y)
(6x2 − 40y)dxdy (84)
1 x1 (y)

It is up to you to check if the answers given by eqs.83 and 84 are equal.

Gaussian Integrals
In mathematics, a Gaussian function, often simply referred to as a Gaussian, is a function of
the form:
(x − b)2

f (x) = ae 2c2 (85)

for arbitrary real a, b, c.


The Gaussian integral, also known as the EulerPoisson integral is the integral of the Gaussian
function over the entire real line:
Z ∞
2
e−x dx (86)
−∞

A family of Gaussian integrals is also can be presented in the form:


Z ∞
2
Ip (λ) = xp e−λx dx Re(p) > 0, Re(λ) ≥ 0 (87)
0

The
Z remarkable
Z fact is that the indefinite integrals from gaussian function, i.e. integrals like
2 2
e−x dx or xp e−λx dx cannot be expressed with elementary functions, but finite integrals
given by eqs.86, 87 can be taken!

Step 1. Let us consider the integrals


Z ∞
2
Y = e−x dx (88)
Z∞∞
2
Y = e−y dy (89)

48
Clearly, Z ∞ Z ∞
2 +y 2 )
Y = 2
e−(x dxdy (90)
−∞ −∞
Note:
Here we deal with so-called multiple integral which is a definite integral of a function of more
than one real variable. Here let’s change the coordinate system from cartesian to polar using
transition formula x = cos φ, y = sin φ and accounting for the fact the elementary surface area or
elementary volume in different coordinate system can be obtained using Jacobian (determinant
of the Jacobian matrix) and the product of corrdinates’ differentials:

dxdy = |Jx,y (ρ, φ)|dρdφ (91)

Then, eq.90 can be written:


Z 2π Z ∞ Z ∞ ∞
2 −ρ2 −ρ2 1 −ρ2
Y = e ρdρdφ = 2π e ρdρ = 2π (−e ) = π (92)
0 0 0 2 0

Consequently, accounting for the fact that Y > 0, obtain:


Z ∞
2 √
Y = e−x dx = π (93)

Step 2:
∞ √
Z
2
Let us consider the integral: e−λx , then making the substitution t = x λ, obtain:
−∞
Z ∞  π 1/2
2
e−λx dx = (94)
−∞ λ
So, taking into account that the integrand (the function under the integral) is even function,
one can write Z ∞
2 1  π 1/2
I0 (λ) = e−λx dx = (95)
0 2 λ
Step 3:

Z ∞ Z ∞
−λx2 1 2
I1 (λ) = xe dx = − (−2λx)e−λx dx =
0 0 2λ
Z ∞
1 −λx2 1 −λx2 ∞ 1
− d(e )=− e = (96)
2λ 0 2λ 0 2λ
Using the same technique, it is possible to calculate
Z ∞ Z ∞ Z ∞
2k+1 −λx2 1 −λx2 1 2
I2k+1 (λ) = x e dx = − 2k
x (−2λx)e dx = − x2k d(e−λx ) =
0 2λ 0 2λ 0

Z inf ty Z inf ty
1 h
2
i 1 2 k 2
− x2k e−λx + (2k) x2k−1 e−λx dx = x2k−1 e−λx dx (97)
2λ 0 2λ 0 λ 0
That means that a recurrent relation has been found:
k kk−1 k! k!
I2k+1 (λ) = I2k−1 = I2k−3 = · · · = k I1 (λ) = k+1 (98)
λ λ λ λ 2λ
49
Step 4:
For the even powers of x:
Z inf ty Z ∞
12k −λx2 2
I2k (λ) = x e dx = − x2k−1 d[e−λx ] =
0 2λ 0
1 2k−1 −λx2 ∞ 2k − 1 ∞ 2k−2 −λx2
Z
− [x e ]0 + x e dx (99)
2λ 2λ 0

A recurrent relation is
2k − 1 2k − 1 2k − 3 1 · 3 · · · (2k − 1)
I2k = I2k−2 = I2k−4 = · · · = I0 (λ)
2λ 2λ 2λ (2λ)k
1 · 3 · · · (2k − 1) 1  π 1/2
= (100)
(2λ)k 2 λ

Summary
Z ∞
2 1 π
I0 = e(−λx ) dx = ( )1/2 (101)
0 2 λ
Consequently, Z ∞
2 π
e(−λx ) dx = ( )1/2 (102)
−∞ λ
Z ∞
2 1
I1 = xe(−λx ) dx = (103)
0 2λ
Consequently, Z ∞
2
xe(−λx ) dx = 0 (104)
−∞

Recurrent relations: Z ∞
2 1 1  π 1/2
I2 = x2 e(−λx ) dx = I0 (λ) = (105)
0 2λ 4λ λ
Consequently, Z ∞
2 (−λx2 ) 1  π 1/2
xe dx = (106)
∞ 2λ λ
Example
Z ∞ Z ∞  π 1/2
−λ(x−x0 )2 2
e dx = e(−λt ) dt = (107)
−∞ −∞ λ
Example
Z ∞ Z ∞ √
(−x2 −y 2 ) (−y 2 ) 2 2)
G(y) = e dx = e e−x dx = e(−y π (108)
−∞ −∞

Example
Z ∞ Z ∞ Z ∞ √
Z ∞
(−x2 −y 2 ) 2
H= G(y)dy = e dxdy = π e(−y ) dy = π (109)
−∞ −∞ −∞ −∞

50
Integrals in physics: line integral, surface integral
Line integral
A line integral is an integral taken along a curve. The terms path integral, curve integral, and
curvilinear integral and contour integral are used as well.

The line can be given in the form of a parametric curve or as and the function can be either
scalar of vector.

Line integral of a scalar function along a parametric curve

Let the C curve along which the integral is evaluated is smooth and is given by the parametric
equations (see the table for parametric presentation of some curves):

x = h(t) y = g(t) a≤t≤b


~r(t) = h(t)~i + g(t)~j a≤t≤b (110)

Then the line integral of a scalar function f (x, y) along C is denoted as


Z
f (x, y)ds
C

where ds is a small part of the curve C. It can be shown that ds is related to the variations of
dt, dx, dy as: s 
2  2
dx dx
ds = + dt (111)
dt dy
Using parametric presentation of a curve with a radius-vector ~r(t) eq.110, it is possible also to
write
ds = k~r0 (t)k (112)

Example: Z
Take the integral xy 4 ds where C is the right half of the circle x2 + y 2 = 16.
C
Solution:
The circle is presented in the parametrization form as x = 4 cos t, y = 4 sin t. For its right half,
−π/2 ≤ t ≤ π/2. Then, taking the require derivatives, obtain:
p
ds = 16 sin2 (t) + 16 cos2 (t)dt = 4t

So, the integral to take is:


Z Z π/2 Z π/2
4 4
xy ds = 4 cos(t)(4 sin(t) )4dt = 4096 cos(t) sin4 (t)dt
C −π/2 −π/2

51
Parametric presentations for some curves in two dimensions
Curve Parametric presentation
Ellipse Counter-Clockwise Clockwise
2
x2
a2
+ yb2 = 1 x = a cos(t) x = a cos(t)
y = b sin(t) y = −b sin(t)
0 ≤ t ≤ 2π 0 ≤ t ≤ 2π
Circle Counter-Clockwise Clockwise
x + y 2 = r2
2
x = r cos(t) x = r cos(t)
y = r sin(t) y = −r sin(t)
0 ≤ t ≤ 2π 0 ≤ t ≤ 2π
y = f (x) x=t
y = f (t)
x = g(y) x = g(t)
y=t
Line segment ~r(t) = (1 − t) · (x0 , y0 , z0 ) + t · (x1 , y1 , z1 ), 0 ≤ t ≤ 1
from (x0 , y0 , z0 ) to (x1 , y1 , z1 ) or
x = (1 − t)x0 + tx1
y = (1 − t)y0 + ty1 , 0 ≤ t ≤ 1
z = (1 − t)z0 + tz1

52
Example Z
Take the integral 4x3 ds over the line shown in the fig.14 Solution:

Figure 14: Path to consider for the integration in the given example
First, a parametrization for each of the curves is needed:

C1 : x = t, y = −1, −2 ≤ t ≤ 0
C2 : x = t, y = t3 − 1, 0≤t≤1
C3 : x = 1, y = t, 0≤t≤2
(113)

Then, the line integral is presented as a sum of 3 integrals:


Z Z Z Z
3 3 3
4x = 4x ds + 4x ds + 4x3 ds
C C1 C2 C3

and each of the integrals are:


Z
3
Z 0 √ 0
4x ds = 4t3 12 + 0dt = t4 −2 = −16
C1 −2
1 Z 1 √
1 1
Z Z Z
p 12
4x3 ds = 3 2 2 2
4t 1 + (3t ) dt = 3 4
4t 1 + 9t dt = sqrt1 + 9t4 d(1 + 9t4 ) = (1 + 9t4 )3/
C2 0 0 9 0 9 3
Z Z 2 p
4x3 ds = 4(1)3 0 + (1)2 dt = 8
C1 0

Example
Let us take the same integral but on a linear path from (−2, −1) and (1, 2). Let’s use radius-
vector for the parametrization of the linear segment.
( ) ( ) ( ) ( ) ( )
−2 1 −2 + 2t t −2 + 3t
~r(t) = (1 − t)(−2, −1) + t(1, 2) = (1 − t) +t = + =
−1 2 −1 + t 2t −1 + 3t
for 0 ≤ t ≤ 1.
That means that individual parametric equations are

x = −2 + 3t, y = −1 + 3t
53
Then
1 √
Z Z
3
4x ds = 4(−2 + 3t)3 9 + 9dt = −21.213 (115)
C 0
Example
Let us take the same integral but on the same linear path but with changing the direction from
(1, 2) to (−2, −1). Note that parametrization changes and becomes
( )
1 − 3t
~r(t) =
2 − 3t

for 0 ≤ t ≤ 1.
The integral becomes:
1 √
Z Z
3
4x ds = 4(1 − 3t)3 9 + 9dt = −21.213 (116)
C 0

So, the change of the direction for the integration path did not change the sign of the integral.

Line integral with respect to the coordinates


Let f (x, y) is given, and parametrization of the path to integrate over is x = x(t), y = y(t)),
a ≤ t ≤ b.
The line integral of f with respect to x is
Z Z b
f (x, y)dx = f (x(t), y(t))x0 (t)dt (117)
C a

The line integral of f with respect to y is


Z Z b
f (x, y)dy = f (x(t), y(t))y 0 (t)dt (118)
C a

These two integral often appear together and can be written as


Z Z Z
P dx + Qdy = P (x, y)dx + Q(x, y)dy (119)
C C C

For these integrals, contrary to the previous case of the integration over the arc length, we
have:
Z Z
f (x, y)dx = − f (x, y)dx
−C
Z ZC
f (x, y)dy = − f (x, y)dy
−C C
Z Z
P dx + Qdy = − P (x, y)dx + Q(x, y)dy (120)
−C C

54
Vector fields
Definition:
A vector field on two (or three) dimensional space is a function F~ that assigns to each point
(x, y) in 2D or (x, y, z) in 3D a two (or three dimensional) vector given by F~ (x, y) in 2D or
F~ (x, y, z) in 3D.
Examples from Physics:
· Velocity field of the liquid (fig.15c)

Figure 15: Illustration to the 2D and 3D vector field

· Magnetic B,~ H~ and Electric fields E,


~ D ~
· Electric current J~
· Gradient of the scalar field which often have
 notion of fluxes:
~ = −κ ∂T ~i + ∂T ~j + ∂T ~k
density of the heat flux ~q = −κ∇T
∂x ∂x ∂x
~
density of the diffusive flux jD = −D∇C~
...
Standard notation for the vector fields::

V~ (x, y, z) = ~iVx (x, y, z) + ~jVy (x, y, z) + ~kVz (x, y, z) = (Vx (x, y, z), Vy (x, y, z), Vz (x, y, z))
F~ = (Fx , Fy , Fz )

Line integral of Vector Fields


Let a vector field is given:

F~ (x, y, z) = P (x, y, z)~i + Q(x, y, z)~j + R(x, y, z)~k

and the smooth curve C is given by the radius vector

~r(t) = x(t)~i + y(t)~j + z(t)~k a≤t≤b

The line integral of F~ along C is


Z Z
~
F · d~r = F~ (~r(t)) · d~r0 (t)dt (121)
C C
55
There are numerous example of the line integrals ofZthe vector fields in physics, e.g. mechanical
Z b
work A = F~ d~r, work of the electric field W = Q ~ r, etc. Illustration to such integration
Ed~
S a
are given in fig.16.

Figure 16: Illustration to the integration of a vector field over a line

Example: Calculate the integral of the vector function F (x,~ y) = 3y~i + 5x~j over the right part
of a circle x2 + y 2 = 16.
Solution
~ y) over the
Using parametrization for the circle, we can present variations of the function F (x,
required path as
~ y) = 3(4 sin(t))~i + 5(4 cos(t))~j
F (x,

and the path is ~r = 4 cos(t)~i + 4 sin(t)~j


Then ~r0 (t) = −4 sin(t)~i + 4 cos(t)~j and the integration to be done is:
Z Z π/2
~
F (x, y)d~r = (12 sin(t)~i + 20 cos(t)~j)(−4 sin(t)~i + 4 cos(t)~j)dt =
C −π/2
Z π/2 Z π/2
2 2
(−48 sin (t) + 80 cos (t))dt = (−48 + 128 cos2 (t))dt =
−π/2 −π/2
Z π/2  π/2
π/2 2 1
−48 t|−π/2 + 128 cos (t)dt = −48t + 128t = 40π (122)
−π/2 2 −π/2

Remark: the primitive of cos2 (t) can be found using integration by parts:
Z Z Z Z
cos (t)dt = cos(t)d sin(t) = cos(t) sin(t) − sin(t)d cos(t) = cos(t) sin(t) + sin2 (t)dt
2

Z Z
= cos(t) sin(t) + (1 − cos (t))dt = cos(t) sin(t) + t − cos2 (t)dt
2

From where
Z Z
1
2 cos2 (t)dt = cos(t) sin(t) + t , i.e. cos2 (t)dt = [cos(t) sin(t) + t]
2

Note that the two-dimensional vector field F~ (x, y) can be drawn schematically since in some
56
Figure 17: Illustration to the integration of a vector field over a line

selected points the vectors can be constructed. Illustration for the field given in the previous
example is shown inZ fig.17 as well as for the integration along the semi-circle. Example:
Take the integral F~ d~r for F~ = xz~i − yz~k and C been the line segment from (−1, 2, 0) to
C
(3, 0, 1).
Solution
Parametrization of the segment (cf. table) using notation in form of vectors:
         
−1
  3
   −1 + t   3t   −1 + 4t 
~r(t) = (1 − t) 2 + t 0 = 2 − 2t + 0 = 2 − 2t for 0 ≤ t ≤ 1

   
   
      
0 1 0 t t
   

that means that we should use presentations for x = 4t − 1, y = 2 − 2t and z = t that gives for
the function F~ = (4t − 1)t~i + (2 − 2t)t~k and d~r = (4~i − 2~j + ~k)dt. Then for the integration:

Z Z 1h i
~
F d~r = (4t2 − t)~i + (2t − 2t2 )~k (4~i − 2~j + ~k)dt =
C
Z 01

18t2 − 6t dt = (6t3 − 3t2 ) 0 1 = 3

(123)
0

Green theorem
Note that without going to parametrization, we can write the differential d~r = ~idx + ~jdy + ~kdz.
Then, presenting the function F~ as

F~ (x, y, z) = P (x, y, z)~i + Q(x, y, z)~i + R(x, y, z)~k

it is possible to write the line integral over a path C of the vector filed F~ in a form of:
Z Z h i h i
~
F (x, y, z)d~r = P~i + Q~i + R~k d ~idx + ~jdy + ~kdz =
C C
Z
P dx + Qdy + Rdz (124)
C
57
i.e. as three integrals, each integral is taken over one variable but along the path C.
Green theorem Let C be a positively oriented, piecewise smooth, simple closed curve in a
plane, and let D be the region bounded by C. If P and Q are functions of (x, y) defined on an
open region containing D and have continuous partial derivatives there, then
I ZZ  
∂Q ∂P
(P dx + Q dy) = − dx dy (125)
C D ∂x ∂y

where the path of integration along C is anticlockwise.

Remark:In mathematics, a positively oriented curve is a planar simple closed curve (that is,
a curve in the plane whose starting point is also the end point and which has no other self-
intersections) such that when traveling on it one always has the curve interior to the left (and
consequently, the curve exterior to the right).

Example
H
Take the integral C y 3 dx − x3 dy where C is a circle of the radius 2 at the origin.
Solution:
Given example satisfies conditions of the Green’s theorem, P = y 3 and Q = −x3 . Then
∂Q/∂x = −3x2 , ∂P/∂y = 3y 2 that gives
I ZZ
3 3
y dx − x dy = − (3x2 + 3y 2 )dxdy (126)
C D

where D is a disk of the radius 1. It is convinient to take this integral in the polar coordinate
for which dxdy = rdrdθ:
Z 1 Z 2π Z 2 2
3πr4
I
3 3 2 3
y dx − x dy = − dr dθ(3r )r = −2π 3r dr = = −24π (127)
C 0 0 0 2 0

Example
H
Take the integral C y 3 dx − x3 dy where C consists of two circles: one of a radius 2and another
of a radius 1 (18).
Solution:
Although the region with hole(s) (multiply connected space) do not satisfy conditions of Green’s
Theorem, the region D between the two circles satisfies them! Another option is to consider
integration over ”4 semi-circles” connected by paths denoted C5 and C6 in fig.18. The integrals
over these paths will be compensated, the rest will be the difference between the integrals taken
over different contours:
Z 2 Z 2π Z 2 2
3πr4
I
3 3 2 3 45π
y dx − x dy = − dr dθ(3r )r = −2π 3r dr = =− (128)
C 1 0 1 2 1 2

58
Figure 18: Integration over a contour consisting of two circles

Surface integrals
Divergence
Given the vector field F~ = P~i + Q~j + R~k the divergence is defined to be
∂P ∂Q ∂R
div F~ = + + (129)
∂x ∂y ∂z
The definition can be written with the nabla operator:

div F~ = ∇
~ · F~ (130)

Example
In hydrodynamics the mass transport equation through a unit volume is written as
∂ρ ~ · V~ = Sm
+ ρ∇
∂t
∂ρ
where Sm is a possible source of the mass. For incompressible fluids = 0, so, if no sources of
∂t
mass exist, the divergence of the velocity field is zero. Compare an illustration for vector fields
given in fig.19a,b. One of the presented field is a divergent-free, while another one Example

Figure 19: Illustration to a divergence-free vector field and a vector field with a sink

In thermal physics a gradient of the temperature field without the heat source is a divergence-
free vector field.
59
Operator of Laplace

Vector form of Green’s Theorem


The vector form of Greens Theorem that uses the divergence is given by
I ZZ
~
F · ~n = div F~ dA (131)
D

Here ~n is a unit normal vector to the curve bounded the area D. The illustration is given in
fig.20

Figure 20: Illustration to the Green’s theorem in a vector form

Surface integral and the Divergence Theorem


Let us remind some facts about integration over a line. We are working with functions of 2
variables (x, y) (or 3 (x, y, z), or more) and we are taking the integral over a line along which
all variables may change. Further, Green’s Theorem relates integration over a closed contour
with integration over the area bounded by this contour.
Similarly, we can work with a function of 3 variables (x, y, z) and perform integration over a
surface along which three variables can varies. Example of such surface in Cartesian space are
cylinder, sphere, any curved surface.
Similar to the previous consideration, we can introduce parametrization of the surface using its
vector presentation, compare :

~r(t) = x(t)~i + y(t)~j + z(t)~k parametrization for a curve


~r(u, v) = x(u, v)~i + y(u, v)~j + z(u, v)~k parametrization for a surface

Further, the function to be integrated f (x, y, z) can also be re-written using this parametriza-
tion, i.e. in the form f (x(u, v)y(u, v), z(u, v)), then the susrface integral
Z
f (x, y, z)dS
S

60
can be reduced to the integration over a some (simpler) region D, i.e. the idea of the integration
of a scalar function over the surface is quite similar to the inegration over a line.

Furthermore, a Divergence theorem for the integration of the vector functions over a
surface, relates integration over a surface and integration over a volume.
Divergence theorem
Let E be a simple solid region and S is the boundary surface of E with positive orientation.
Let F~ be a vector field whose components have continuous first order partial derivatives. Then,
ZZ ZZZ
~ ~
F · dS = div F~ dV (132)
S E

61
Differential equations
Classification of the differential equations:
1. With respect to number of variables: ordinary or partial
· ordinary differential equation (ODE) is a differential equation containing one or more
functions of one independent variable and its derivatives
· partial differential equations are those that contain unknown multivariable functions and
their partial derivatives
2.With respect to the order of the highes derivative: first order, second order, etc....
3. Linear or non-linear:
Linear ordinary differential equations are of the form

Ly = f or, more precisely L[y(t)] = f (t) (133)

where the differential operator L is a linear operator. A linear differential operator, involving
the n-th derivative, Ln may be considered to be of the form

dn y dn−1 y dy
Ln [y(t)] ≡ n
+ A1 (t) n−1
+ · · · + An−1 (t) + An (t)y (134)
dt dt dt
Particular solution of a differential equation
Particular solution for a differential equation is any function which satisfies differential equation
on the required interval.

General solution of a differential equation


General solution of a differential equation is a solution which contains all constants of the
integration.
Otherwise it is sometimes said that particular solution is obtained from the general solution
for particular values of the constants.

Initial Condition(s)
Initial conditions are given values of the function and/or its derivative(s) at specific points.
Initial Condition(s) allows us to choose one solution from the general solution.
Initial Value Problem
An Initial Value Problem (or IVP) is a differential equation along with an appropriate number
of initial conditions.
Example
dy
y0 = = f (x)
dx
Solution: Z
y = f (x)dx + C

62
Example

00d2 y
y = 2 = −g
dt
y(0) = 3y 0 (0) = V0

Solution:
Z
0
y = gdt + C1 = gt + C1
gt2
Z
y= (gt + C1 )dt = + C1 t + C2
2
Using initial conditions:

y(0) = C2 => C2 = 3
y 0 (0) = C1 => C1 = V0

Finally:
gt2
y(t) = + V0 t + 3
2

Ordinary differential equations of first order: first order ODE


Methods to solve:
Separation of variables
Substitutions
Exact equations
Inexact equations integrating factors
Variation of parameters
Power series solutions

Separation of variables

Let equation has the form

y 0 = f (x) · g(y) Then:


dy
= f (x) · g(y) for g(y 6= 0)
dx Z Z
dy
= f (x)dx
g(y)
Note that the dif.eq. in the form
Z Z
f1 (x) g2 (y)
f1 (x) · f2 (y)dx + g1 (x) · g2 (y)dy = 0 dx + dy = C
g1 (x) f2 (y)
can also be re-written as
 
f1 (x) g2 (y)
g1 (x) · f2 (y) dx + dy = 0
g1 (x) f2 (y)
63
That means that besides the general solution of eq.135 another solution can exist for the case
g1 (x) · f2 (y) = 0.
Example

x2 y 0 + y − 1 = 0

Solution:
dy dx
= 2 for x2 6= 0 andy 6= 1
1−y x
1
− ln |1 − y| = − + C3
x
|1 − y| = e1/x−C3
= Ce 1/x
where C = e−C3 , C > 0

Finally:
y = 1 − Ce1/x for C > 0 and C < 0 (135)

Yet, this solution has probably to be updated since it was obtained under condition y − 1 6= 0
and x 6= 0. Actually, eq.135 includes solution y = 1 if C = 0. But the case x = 0 has to be
considered separately, so, finally we have:

y = 1 − Ce1/x for x 6= 0 and all C


y = 1 for x = 0 as a special solution

Problem
Find the general solution for
xy 0 = 2y + y 0 (136)

and find a particular solution for this problem with x0 = 2, y0 = 3. Solution:

y 0 (x − 1) = 2y
dy dx
=2 for x 6= 1 and y 6= 0
y x−1
ln |x| = 2 ln |x − 1| + C = ln(x − 1)2 + C
|y| = C(x − 1)2 for C > 0 and C < 0

No special solution exists for this case since x = 1 implies y = 0 that can be obtained if C = 0.
So, the last formula is for all values of C.
A particular solution: if for x = 2, y = 3, we can find a particular value for c and get
finally:y = 3(x − 1)2 .

Problem with illustration for the solution in an implicit form


Find the general solution for
yy 0 = ey

64
First, we shall separate the variable and then we will integrate the equation:
dy
y = ey
dx
ye−y = dx
Z
Note that ye−y dy = −ye−y − e−y
−ye−y − e−y = x + C
(y + 1)e−y = −x + C

Solution given by the last eqution has an implicit form. Soltion exist for all x and y 6= 0.

Substitutions

Case 1: for
y 0 = f (ax + by + c) (137)

use the substitution z = ax + by + c and note hat according to initial equation 137, y 0 = f (z)
Then:
dz dy
=a+b
dx dx
z 0 = a + by 0 = a + bf (z)
dz
= dx
a + bf (z)
Z Z
dz
= dx for a + bf (z) 6= 0 (138)
a + bf (z)

Case 2: for

y 0 = f (y/x) (139)

use the substitution z = y/x and note hat according to initial equation 139 y 0 = f (z). Then

y = zx
y0 = z0x + z
z 0 x + z = f (z)
Z Z
dz dx
= for f (z) − z 6= 0
f (z) − z x
(140)

Example
0 y y
y = + tan
x x

65
Solution: take v = y/x. Then:

y0 = v0x + x
0
Zv x + v = v + tan v
cos v
dv = ln x + C
sin v
ln(sin v) = ln x + C
v = arcsin(Cx)
y = x arcsin(Cx) (141)

No solution fox x = 0: y 0 is not defined for such points.

Exact first order ordinary differential equation

Total differential of a function f (x, y) is


∂f ∂f
df (x, y) = dx + dy
∂x ∂y
An exact first order ordinary differential equation is one of the form
∂A ∂B
A(x, y)dx + B(x, y)dy = 0 where = (142)
∂y ∂x
Because in this case a function U (x, y) exists such that dU (x, y) = A(x, y)dx + B(x, y)dy for
which ∂A/∂y = ∂B/∂x = ∂ 2 U/∂y∂x.
Note that:

dU (x, y) = A(x, y)dx + B(x, y)dy = 0


Z
dU (x, y) = U (x, y) = C

On the other hand:


∂U (x, y)
= A(x, y)
Z ∂x Z
∂U (x, y)
U (x, y) = dx + F (y) = A(x, y)dx + F (y) = C
∂x
(143)

Then, taking a partial derivative of the last equation over y, and accounting for the fact that
∂U
= B(x, y) obtain:
∂y Z
∂ dF (y)
A(x, y)dx + =0
∂y dy
which allows us to find F (y).
Example
dy
x + 3x + y = 0 (144)
dx
66
The given equation can be re-written as (3x + y)dx + xdy = 0, A(x, y) = 3x + y, B = x. If
we chack the condition:∂A/∂y = ∂B/∂x = 1, so the equation is exact. Let us find a function
U(x,y):
3x2
Z
U (x, y) = (A(x, y)dx + F (y) = + xy + F (y) = C0 (145)
2
Taking into account that ∂U (x, y)/∂y = B(x, y) and taking the partial derivative of 145 on y,
obtain:
dF (y)
x+ =x
dy
dF (y)
=0
dy
F (y) = C1 (146)

So, finally:
3x2
U (x, y) = + xy + C1 = C0
2
or
3x2
U (x, y) = + xy = C
2

Inexact equations: integrating factors

Let the First order ODE is given as

A(x, y)dx + B(x, y)dy = 0

but ∂A/∂y 6= ∂B/∂x. So, it is not a differential of a certain function. The idea is to find the
integrating factor µ(x, y) such that:
∂(µA) ∂(µB)
=
∂y ∂x
Yet, find the integrating factor µ which would be a function of both variable, x and y, is
quite complicated and no general procedure exists.
However, it is possible to find µ as a function of only x or only y, as explained below.
Let an inexact differential equation is given:
∂M ∂N
M (x, y)dx + N (x, y)dy = 0, 6=
∂y ∂x
and an integrating factor depending only on x exists, that means that
∂(µ(x)M ) ∂(µ(x)N )
=
∂y ∂x
since µ(x) does not depend on y, obtain:
∂M ∂N dµ(x)
µ(x) = µ(x) +N
∂y ∂x dx
 
∂M ∂N dµ(x)
µ(x) − =N
∂y ∂x dx
 
dµ(x) 1 ∂M ∂N 1
= −
dx µ(x) ∂y ∂x N
67
If on the left side the function depend only on x, therefore, on the right side we also should
have a function depending only on x. Let us denote the right side as g(x). Then:

dµ(x) 1
= g(x)
dx µ(x)
Z
ln(µ(x)) = g(x)dx
Z
g(x)dx ∂M

∂N 1
µ(x) = e , where g(x) = − (147)
∂y ∂x N

Remarks:
1. The role of x and y can be reversed, then we have to look for a function h(y):

 
∂M ∂N 1
h(y) = −
∂y ∂x M
2. For the first order ODE of the type
dy
+ a(x)y = b(x)
dx
the integrating factor is
Z
a(x)dx
µ=e

Example

ydx + (2x − yey )dy = 0

here M = y, N = (2x − yey ). Then, find g(x):


1−2 1
g(x) = y
= y
2x − ye ye − 2x
but the right side is a function of y, i.e. we have to try another approach. Let us find h(y):
1−2 1
h(y) = =
y y
Then Z
dy/y
µ(y) = e =y

Multiply initial equation by µ(y):

y 2 dx + (2xy − y 2 ey )dy = 0

68
This equation can be solved using the method of solution for exact equation. Let us find U (x, y):
Z
U (x, y) = y 2 dx + F (y) = y 2 x + F (y) = C0
dF (y)
2yx + = 2xy − y 2 ey
dy
dF = −y 2 eydy => F (y) = −ey (y 2 − 2y + 2) + C1
U (x, y) = C0 = y 2 x − ey (y 2 − 2y + 2) + C1

Finally: y 2 x − ey (y 2 − 2y + 2) = C
The last step is to verify that with taking the appropriate derivative we really get the initial
equation.

Variation of constant parameter

Let a differential equation


dy
+ a(x)y = b(x)
dx
is given. The method of variation of a constant parameter consists of 2 steps:
dy
1. Find a general solution of the corresponding homogeneous equation + a(x)y = 0. Which
dx
will have a form
Z
a(x)dx
y(x) = Ce

2. Assume that the constant C depends on (x) and plug the obtained solution into the differ-
ential equation. Find C(x)

69

Das könnte Ihnen auch gefallen