AMM 409: Partial Differetial Equation I

Instruction Hours: 45
Pre-requisites: AMM 316
Purpose
To introduce students to formulation and solutions of elementary first order partial differential
equations

Expected Learning Outcomes

By the end of the course the learner should be able to:
i) Form a partial differential equations arising from natural phenomena
ii) Use different methods to solve first order partial differential equation of first degree
iii) Use methods of Cauchy, Charpit and Jakobi in solving non-linear partial differential
equation of first order.

Course Content
Surfaces and curves in three dimension; Simultaneous partial differential equation of the first
dx dy
order; Method of solutions /p = /Q = dz/R ;; Orthogonal trajectory and system of curves on a
surface; Partial differential equation: Linear; Semi-linear and; Quasi-linear equations of the first
order; Integral surfaces passing through a given system curve; Use methods of Cauchy, Charpit
and Jakobi in solving non-linear partial differential equation of first order

Teaching / Learning Methodologies

Lectures; Tutorials; Class discussion

Instructional Materials and Equipment

Handouts; Chalk board

Course Assessment
Examination - 70%; Continuous Assessments (Exercises and Tests) - 30%; Total - 100%

1
Recommended Text Books
1. Kevorkian (2006); Partial Differential Equations
2. Shiing-shenChern (2006); Partial Differential Equations; Springer
3. Devaney R. L (2003); Differential Equations; Brooks Publishing

Text Books for further Reading

1. Paul (2006); Differential Equations; Brook and Cole Publications
2. Kneysting Erwin (1993); Advanced Engineering Mathematics; McGraw- Hill London
3. Duchateau Paul and David W. Zachmann (1986); Partial Differential Equations;
London: McGraw Hill

2
Partial Differential Equations of the First Order
Introduction
Partial differential equations are those which contain one or more partial derivatives. They must,
therefore, involve at least two independent variables. The order of a partial differential equation
is that of the derivative of highest order in the equation. For example, considering 𝑧 as dependent
variable and 𝑥, 𝑦 as independent variables,
𝜕𝑧 𝜕𝑧
(𝑎)𝑥 +𝑦 = 𝑧 𝑜𝑟 𝑥𝑝 + 𝑦𝑞 = 𝑧 𝑖𝑠 𝑜𝑓 𝑜𝑟𝑑𝑒𝑟 𝑜𝑛𝑒, 𝑎𝑛𝑑
𝜕𝑥 𝜕𝑦
𝜕 2𝑧 𝜕 2𝑧 𝜕 2𝑧
(𝑏) + 3 + = 0 𝑜𝑟 𝑟 + 3𝑠 + 𝑡 = 0 𝑖𝑠 𝑜𝑓 𝑜𝑟𝑑𝑒𝑟 𝑡𝑤𝑜.
𝜕𝑥 2 𝜕𝑥𝜕𝑦 𝜕𝑦 2
𝜕𝑧 𝜕𝑧 𝜕2𝑧
In writing (a) and (b), use has been made of the standard notation 𝑝 = 𝜕𝑥 , 𝑞 = 𝜕𝑦 , 𝑟 = 𝜕𝑥 2,

𝜕 2𝑧 𝜕 2𝑧
𝑠= 𝑎𝑛𝑑 𝑡 = 2 .
𝜕𝑥𝜕𝑦 𝜕𝑦
Partial differential equations of first order occur in many practical situations in Science and
Engineering. Such situations include Brownian motion, noise in communication systems,
radioactive disintegration, population growth and in many problems dealing with telephone
traffic, traffic flow along a highway, gas dynamics and so on.
The study of partial differential equations of first order is essential to understand the nature of
solutions, and forms a guide to find the solutions of higher order partial differential equations.
A first order partial differential equation (usually denoted by PDE) in two independent
variables 𝑥, 𝑦 and one unknown 𝑧, also called dependent variable, is an equation of the form
𝜕𝑧 𝜕𝑧
𝐹 (𝑥, 𝑦, 𝑧, , ) = 0 … (1)
𝜕𝑥 𝜕𝑦
𝜕𝑧 𝜕𝑧
e.g. 𝑦 𝜕𝑥 + 𝑥 𝜕𝑦 + 2𝑥 − 𝑦 + 𝑧 = 0.
𝜕𝑧 𝜕𝑧
Introducing the notation 𝑝 = 𝜕𝑥 , 𝑞 = 𝜕𝑦 … (2)

Equation (1) can be written in symbolic form as 𝐹(𝑥, 𝑦, 𝑧, 𝑝, 𝑞) = 0 … (3)

A solution of (1) is a function 𝑧 = 𝑓(𝑥, 𝑦) which satisfies equation (1) or (3).
For example, the function 𝑧 = 𝑎𝑥 + 𝑏𝑦 + 𝑎𝑏, 𝑤ℎ𝑒𝑟𝑒 𝑎 𝑎𝑛𝑑 𝑏 are arbitrary constants is a
𝜕𝑧 𝜕𝑧 𝜕𝑧 𝜕𝑧
solution of the first order PDE 𝑥 𝜕𝑥 + 𝑦 𝜕𝑦 + 𝜕𝑥 𝜕𝑦 − 𝑧 = 0 𝑜𝑟 𝑝𝑥 + 𝑞𝑦 + 𝑝𝑞 − 𝑧 = 0.

This is clearly so since from the given function 𝑧 = 𝑎𝑥 + 𝑏𝑦 + 𝑎𝑏,

3
 𝜕𝑧 𝜕𝑧
= 𝑎, =𝑏
𝜕𝑥 𝜕𝑦
Substitution in the given PDE gives 𝑥𝑎 + 𝑦𝑏 + 𝑎𝑏 − (𝑎𝑥 + 𝑏𝑦 + 𝑎𝑏) = 0, thus satisfying it.

Classification

We classify the PDE of first order depending upon the form of the function F. An equation of the
𝜕𝑧 𝜕𝑧
form 𝑃(𝑥, 𝑦, 𝑧) 𝜕𝑥 + 𝑄(𝑥, 𝑦, 𝑧) 𝜕𝑥 = 𝑅(𝑥, 𝑦, 𝑧) … (4)

𝜕𝑧 𝜕𝑧
is a quasi – linear PDE of first order, if the derivatives 𝜕𝑥 and 𝜕𝑦 that appear in the function F are
linear, while the coefficients P, Q and R depend on the independent variables 𝑥, 𝑦 and also on the
dependent variable 𝑧.
𝜕𝑧 𝜕𝑧
Similarly, an equation of the form 𝑃(𝑥, 𝑦) 𝜕𝑥 + 𝑄(𝑥, 𝑦) 𝜕𝑦 = 𝑅(𝑥, 𝑦, 𝑧) … (5)

is called almost linear PDE of first order, if the coefficients P and Q are functions of the
independent variables only.
𝜕𝑧 𝜕𝑧
An equation of the form 𝑎(𝑥. 𝑦) 𝜕𝑥 + 𝑏(𝑥, 𝑦) 𝜕𝑦 + 𝑐(𝑥, 𝑦)𝑧 = 𝑑(𝑥. 𝑦) … (6)

𝜕𝑧 𝜕𝑧
is called a linear PDE of first order, if the function F is linear in 𝜕𝑥 , 𝜕𝑦 and 𝑧, while the
coefficients 𝑎, 𝑏, 𝑐 and 𝑑 depend only on the independent variables 𝑥 and 𝑦.

An equation which does not fit into any of the above categories is called non – linear. For
example,
𝜕𝑧 𝜕𝑧
(i) 𝑥 𝜕𝑥 + 𝑦 𝜕𝑦 = 𝑛𝑧, is a linear PDE of first order.

𝜕𝑧 𝜕𝑧
(ii) 𝑥 𝜕𝑥 + 𝑦 𝜕𝑦 = 𝑧 2 ,is an almost linear PDE of first order.

𝜕𝑧 𝜕𝑧
(iii) 𝑃(𝑧) 𝜕𝑥 + 𝜕𝑦 = 0, is a quasi – linear PDE of first order.

𝜕𝑧 𝜕𝑧
(iv) (𝜕𝑥)2 + (𝜕𝑦)2 = 1, is a non –linear PDE of first order.

Before discussing various methods for finding the solutions of first order PDEs, we shall review
some of the basic concepts and definitions needed from calculus.

Surfaces and Normals

Let 𝐹(𝑥, 𝑦, 𝑧) be a function in 𝑅 3 . Then the vector – valued function grad F can be written as

4
 𝜕𝐹 𝜕𝐹 𝜕𝐹 𝜕𝐹 𝜕𝐹 𝜕𝐹
𝑔𝑟𝑎𝑑 𝐹 = 𝒊 +𝒋 +𝒌 =( , , ) … (7)
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑧

If we assume that the partial derivatives of F do not vanish simultaneously at any point, then the
set of points (𝑥, 𝑦, 𝑧) satisfying the equation 𝐹(𝑥, 𝑦, 𝑧) = 𝑐 … (8)

is a surface for some constant 𝑐.

This surface denoted by 𝑆𝑐 is called a level surface of F. If (𝑥0 , 𝑦0 , 𝑧0 ) is a given point, then by
taking 𝐹(𝑥0 , 𝑦0 , 𝑧0 ) = 𝑐, we get an equation of the form 𝐹(𝑥, 𝑦, 𝑧) = 𝐹(𝑥0 , 𝑦0 , 𝑧0 ) … (9)

which represents a surface passing through the point (𝑥0 , 𝑦0 , 𝑧0 ).

Here, equation (8) represents a one – parameter family of surface in𝑅 3 . For example,

𝑥 2 + 𝑦 2 + 𝑧 2 = 32 is a sphere of radius 3 units while the equation 𝑥 2 + 𝑦 2 + 𝑧 2 = 𝑐 represents

a family of spheres whose radius is √𝑐.

The value of grad F is a vector, normal to the level surface.

Now, one may ask, if it is possible to solve equation (8) for z in terms of x and y. To answer this
question, let us consider a set of relations of the form

𝑥 = 𝑓1 (𝑢, 𝑣), 𝑦 = 𝑓2 (𝑢, 𝑣), 𝑧 = 𝑓3 (𝑢, 𝑣) … (10)

Here for every pair of values of u and v, we will have three numbers x, y and z, which represents
a point in space. However, it may be noted that, every point in space need not correspond to a u
and v. But, if the Jacobian

𝜕(𝑓1 , 𝑓2 )
≠ 0 … (11)
𝜕(𝑢, 𝑣)

then, the first two equations of (10) can be solved and u and v, can be expressed as functions of x
and y like 𝑢 = 𝛾(𝑥, 𝑦), 𝑣 = 𝜇(𝑥, 𝑦).

Thus, u and v are obtained once x and y are known, and the third relation of equation (10) gives
the value of z in the form 𝑧 = 𝑓3 (𝛾(𝑥, 𝑦), 𝜇(𝑥, 𝑦)) … (12)

This is, of course, a functional relation between the coordinates x, y and z as in equation (8).

Hence, any point (x, y, z) obtained from equation (10) always lies on a fixed surface.

Equations (10) are also called parametric equations of a surface. It may be noted that the
parametric equation of a surface need not be unique, which can be seen from the following
example:

5
The parametric equations

𝑥 = 𝑟𝑠𝑖𝑛𝜃 cos ∅, 𝑦 = 𝑟 sin 𝜃 sin ∅, 𝑧 = 𝑟 cos 𝜃

and

(1 − ∅2 ) (1 − ∅2 ) 2𝑟∅
𝑥=𝑟 2
cos 𝜃 , 𝑦 = 𝑟 2
sin 𝜃 , 𝑧 =
(1 + ∅ ) (1 + ∅ ) 1 + ∅2

both represent the same surface 𝑥 2 + 𝑦 2 + 𝑧 2 = 𝑟 2 , which is a sphere where r is a constant.

Examples

1.Show that 𝛁∅ is a vector perpendicular to the surface ∅(𝑥, 𝑦, 𝑧) = 𝑐 where c is a constant.

Let 𝒓 = 𝑥𝒊 + 𝑦𝒋 + 𝑧𝒌 be the position vector to any point 𝑃(𝑥, 𝑦, 𝑧) on the surface. Then

𝑑𝒓 = 𝑑𝑥𝒊 + 𝑑𝑦𝒋 + 𝑑𝑧𝒌 lies in the tangent plane to the surface at P.

𝜕∅ 𝜕∅ 𝜕∅ 𝜕∅ 𝜕∅ 𝜕∅
But 𝑑∅ = 𝜕𝑥 𝑑𝑥 + 𝜕𝑦 𝑑𝑦 + 𝜕𝑧 𝑑𝑧 = 0 or (𝜕𝑥 𝒊 + 𝜕𝑦 𝒋 + 𝜕𝑧 𝒌) . (𝑑𝑥𝒊 + 𝑑𝑦𝒋 + 𝑑𝑧𝒌) = 0

i.e. 𝛁∅. 𝑑𝑟 = 0 so that 𝛁∅ is perpendicular to 𝑑𝒓 and therefore to the surface.

2.Find a unit vector normal to the surface 𝑥 2 𝑦 + 2𝑥𝑧 = 4 at the point (2, −2,3).

𝛁(𝑥 2 𝑦 + 2𝑥𝑧) = (2𝑥𝑦 + 2𝑧)𝒊 + 𝑥 2 𝒋 + 2𝑥𝒌 = −2𝒊 + 4𝒋 + 4𝒌 at the point (2, −2,3).
−2𝒊+4𝒋+4𝒌 1 2 2
Then a unit normal to the surface = = − 3 𝒊 + 3 𝒋 + 3 𝒌.
√(−2)2 +(4)2 +(4)2

3.Find an equation for the tangent plane to the surface 2𝑥𝑧 2 − 3𝑥𝑦 − 4𝑥 = 7 at the point
(1, −1,2).

𝛁(2𝑥𝑧 2 − 3𝑥𝑦 − 4𝑥) = (2𝑥 2 − 3𝑦 − 4)𝒊 − 3𝑥𝒋 + 4𝑥𝑧𝒌

Then a normal to the surface at the point (1, −1,2) is7𝒊 − 3𝒋 + 8𝒌.

The equation of a plane passing through a point whose position vector is 𝒓0 and which is
perpendicular to the normal N is (r – 𝒓0 ) . N = 0. Then the required equation is

[(𝑥𝒊 + 𝑦𝒋 + 𝑧𝒌) − (𝒊 − 𝒋 + 2𝒌)]. (7𝒊 − 3𝒋 + 8𝒌) = 0 𝑜𝑟

7(𝑥 − 1) − 3(𝑦 + 1) + 8(𝑧 − 2) = 0.

Curves and their Tangents

A curve in three – dimensional space 𝑅 3 can be described in terms of parametric equations.

6
Suppose 𝑟⃗ denotes the position vector of a point on a curve C, then the vector equation of C may
be written as

𝑟⃗ = 𝐹⃗ (𝑡) 𝑓𝑜𝑟 𝑡 ∈ 𝐼,

where I is some interval on the real axis. In component form, the above equation can be written
as

𝑥 = 𝑓1 (𝑡), 𝑦 = 𝑓2 (𝑡), 𝑧 = 𝑓3 (𝑡)

where 𝑟⃗ = (𝑥, 𝑦, 𝑧) and 𝐹⃗ (𝑡) = [𝑓1 (𝑡), 𝑓2 (𝑡), 𝑓3 (𝑡)].

Further, we assume that

𝑑𝑓1 (𝑡) 𝑑𝑓2 (𝑡) 𝑑𝑓3 (𝑡)

( , , ) ≠ (0,0,0)
𝑑𝑡 𝑑𝑡 𝑑𝑡
This non – vanishing vector is tangent to the curve C at the point (𝑥, 𝑦, 𝑧) or at
[𝑓1 (𝑡), 𝑓2 (𝑡), 𝑓3 (𝑡)] of the curve C.

Another way of describing a curve in three – dimensional space is by using the fact that the
intersection of two surfaces gives rise to a curve.

Let

𝐹1 (𝑥, 𝑦, 𝑧) = 𝑐1
}
𝐹2 (𝑥, 𝑦, 𝑧) = 𝑐2

be two surfaces. Their intersection, if not empty, is always a curve, provided grad 𝐹1 and grad 𝐹2

are not collinear at any point of 𝑅 3 . In other words, the intersection of two surfaces is a curve if

grad 𝐹1 (𝑥, 𝑦, 𝑧) 𝑋 grad 𝐹2 (𝑥, 𝑦, 𝑧) ≠ (0,0,0)𝑓𝑜𝑟 𝑒𝑣𝑒𝑟𝑦 (𝑥, 𝑦, 𝑧) ∈ 𝑅 3

For various values of 𝑐1 and 𝑐2 , the equation describes different curves. The totality of these
curves is called a two parameter family of curves. Here, 𝑐1 and 𝑐2 are referred to as parameters
of this family.

Example
𝜋
Find the tangent vector at (0,1, 2 ) to the helix described by the equation

𝑥 = cos 𝑡, 𝑦 = sin 𝑡, 𝑧 = 𝑡 , 𝑡 ∈ 𝐼 𝑖𝑛 𝑅

Solution

The tangent vector to the helix at (𝑥, 𝑦, 𝑧) is

7
 𝑑𝑥 𝑑𝑦 𝑑𝑧
( , , ) = (− sin 𝑡, cos 𝑡, 1).
𝑑𝑡 𝑑𝑡 𝑑𝑡
𝜋 𝜋 𝜋
We observe that the point (0,1, 2 ) corresponds to 𝑡 = 2 . At this point(0,1, 2 ), the tangent vector
to the given helix is (−1, 0, 1).

FORMATION OF PARTIAL DIFFERENTIAL EQUATIONS

Formation from real life situations

Functions in real world mostly depend on many parameters. For example, Temperature is a
function of several parameters such as Time, Latitude, Longitude, Altitude etc…

Partial Differential Equations (PDEs) are used to precisely model this type of functions with
many variable parameters.

Suppose we want to study temperature(𝜃) variation with respect to time(t), i.e how likely the
𝜕𝜃
temperature will vary with respect to time, we take derivative w. r. t time 𝜕𝑡 .

Similarly if we want to study temperature variation with respect to altitude, we take derivative
𝜕𝜃
with respect to altitude 𝜕ℎ, and so on. Using the physical conditions provided, the information
given can be expressed in terms of a Partial differential equation.

In general therefore, in real life, partial differential equations may arise in connection with
geometrical and physical problems.

Besides, partial differential equations may be derived or obtained by the elimination of arbitrary
constants from a given relation between the variables and by the elimination of arbitrary
functions of the variables. This is demonstrated in each case with examples as shown below.

Deriving a Partial differential equation by elimination of arbitrary constants

Partial differential equations may be obtained by elimination of arbitrary constants.

Examples

1. Eliminate the arbitrary constants 𝑎 and 𝑏 from 𝑧 = 𝑎𝑥 + 𝑏𝑦 + 𝑎𝑏.

𝜕𝑧 𝜕𝑧
Differentiating partially with respect to 𝑥 and 𝑦, we have 𝜕𝑥 = 𝑝 = 𝑎 and 𝜕𝑦 = 𝑞 = 𝑏.

Substituting for 𝑎 and 𝑏 in the given relation we obtain 𝑧 = 𝑝𝑥 + 𝑞𝑦 + 𝑝𝑞, or

𝜕𝑧 𝜕𝑧 𝜕𝑧 𝜕𝑧
𝑧=𝑥 +𝑦 + 𝑖. 𝑒.
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑦

8
 𝜕𝑧 𝜕𝑧 𝜕𝑧 𝜕𝑧
𝑥 +𝑦 + − 𝑧 = 0 𝑜𝑟 𝑝𝑥 + 𝑞𝑦 + 𝑝𝑞 − 𝑧 = 0,
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑦

which is the corresponding first order partial differential equation.

If there is only one arbitrary constant, two possibilities arise.

2. Eliminate the arbitrary constant 𝑎 from the relation 𝑧 = 𝑎(𝑥 + 𝑦).

Differentiating partially with respect to 𝑥 gives 𝑝 = 𝑎, so that the partial differential equation

𝜕𝑧
𝑧 = 𝑝(𝑥 + 𝑦)𝑜𝑟 (𝑥 + 𝑦) −𝑧 =0
𝜕𝑥
is obtained. Similarly, partial differentiation with respect to 𝑦 gives 𝑞 = 𝑎 and the corresponding
partial differential equation

𝜕𝑧
𝑧 = 𝑞(𝑥 + 𝑦)𝑜𝑟 (𝑥 + 𝑦) − 𝑧 = 0.
𝜕𝑦

Both first order partial differential equations are admissible and their solution is the relation 𝑧 =
𝑎(𝑥 + 𝑦).

3.Eliminate the arbitrary constants 𝑎 and 𝑏 from 𝑧 = 𝑎𝑥 2 + 𝑏𝑦 2 + 𝑎𝑏.

𝜕𝑧 𝜕𝑧
Differentiating partially with respect to 𝑥 and 𝑦, we have 𝜕𝑥 = 𝑝 = 2𝑎𝑥 and 𝜕𝑦 = 𝑞 = 2𝑏𝑦.

Solving for 𝑎 and 𝑏 from these equations and substituting in the given relation we obtain

1𝑝 2 1𝑞 1𝑝 1𝑞
𝑧=( ) 𝑥 + ( ) 𝑦 2 + ( ) ( ) 𝑜𝑟 𝑝𝑞 + 2𝑝𝑥 2 𝑦 + 2𝑞𝑥𝑦 2 = 4𝑥𝑦𝑧 𝑖. 𝑒
2𝑥 2𝑦 2𝑥 2𝑦

𝜕𝑧 𝜕𝑧 𝜕𝑧 𝜕𝑧
+ 2𝑥 2 𝑦 + 2𝑥𝑦 2 = 4𝑥𝑦𝑧,
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑦

which is a partial differential equation of order one.

If the number of arbitrary constants to be eliminated exceeds the number of independent

variables, the resulting partial differential equation(or equations) is usually of order higher than
the first.

4.Eliminate ‘a’, ‘b’ and ‘c’ from the relation 𝑧 = 𝑎𝑥 + 𝑏𝑦 + 𝑐𝑥𝑦.

Differentiating partially with respect to 𝑥 and 𝑦, we have (𝑖)𝑝 = 𝑎 + 𝑐𝑦 𝑎𝑛𝑑 (𝑖𝑖)𝑞 = 𝑏 + 𝑐𝑥.

These, together with the given equation, are not sufficient for the elimination of three constants.

Differentiating (i) partially with respect to x, we have

9
 𝜕𝑝 𝜕 2 𝑧
= = 𝑟 = 0, 𝑎 𝑝𝑎𝑟𝑡𝑖𝑎𝑙 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡𝑖𝑎𝑙 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑜𝑟𝑑𝑒𝑟 𝑡𝑤𝑜.
𝜕𝑥 𝜕𝑥 2
Differentiating (ii) partially with respect to y, we have

𝜕𝑞 𝜕 2 𝑧
= = 𝑡 = 0, 𝑎 𝑝𝑎𝑟𝑡𝑖𝑎𝑙 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡𝑖𝑎𝑙 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑜𝑟𝑑𝑒𝑟 𝑡𝑤𝑜.
𝜕𝑦 𝜕𝑦 2

Differentiating (i) partially with respect to y or (ii) with respect to x, we obtain

𝜕𝑝 𝜕𝑞 𝜕 2𝑧
= = = 𝑠 = 𝑐.
𝜕𝑦 𝜕𝑥 𝜕𝑥𝜕𝑦

From (i), 𝑝 = 𝑎 + 𝑠𝑦 𝑎𝑛𝑑 𝑎 = 𝑝 − 𝑠𝑦;

From (ii), 𝑏 = 𝑞 − 𝑠𝑥.

Substituting for ‘a’, ‘b’ and ‘c’ in the equation 𝑧 = 𝑎𝑥 + 𝑏𝑦 + 𝑐𝑥𝑦, we obtain

𝑧 = (𝑝 − 𝑠𝑦)𝑥 + (𝑞 − 𝑠𝑥)𝑦 + 𝑠𝑥𝑦 = 𝑝𝑥 + 𝑞𝑦 − 𝑠𝑥𝑦 𝑜𝑟

𝑝𝑥 + 𝑞𝑦 − 𝑠𝑥𝑦 − 𝑧 = 0, 𝑎 𝑝𝑎𝑟𝑡𝑖𝑎𝑙 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡𝑖𝑎𝑙 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑜𝑟𝑑𝑒𝑟 𝑡𝑤𝑜.

Thus, we have three partial differential equations 𝑟 = 𝑜, 𝑡 = 0, 𝑝𝑥 + 𝑞𝑦 − 𝑠𝑥𝑦 − 𝑧 = 0 of the

same order (second order), associated with the given relation i.e.

𝜕 2𝑧
= 0;
𝜕𝑥 2
𝜕 2𝑧
= 0;
𝜕𝑦 2

𝜕𝑧 𝜕𝑧 𝜕 2𝑧
𝑥 +𝑦 − 𝑥𝑦 − 𝑧 = 0.
𝜕𝑥 𝜕𝑦 𝜕𝑥𝜕𝑦

All three second order PDEs are applicable.

Exercise

1.Eliminate the arbitrary constants ‘a’ and ‘b’ from the following equation to derive the
corresponding partial differential equation.

𝑧 = (𝑥 − 𝑎)2 + (𝑦 − 𝑏)2 .

Solution

Differentiating partially with respect to x and y, we have

10
 𝜕𝑧 𝜕𝑧
= 2(𝑥 − 𝑎); = 2(𝑦 − 𝑏).
𝜕𝑥 𝜕𝑦
1 𝜕𝑧 1 𝜕𝑧
Therefore, 𝑥 − 𝑎 = 2 𝜕𝑥 ; 𝑦 − 𝑏 = 2 𝜕𝑦.

1 𝜕𝑧 1 𝜕𝑧 1 𝜕𝑧 1 𝜕𝑧
Substitution in the given equation gives 𝑧 = (2 𝜕𝑥)2 + (2 𝜕𝑦)2 𝑜𝑟 𝑧 = 4 (𝜕𝑥)2 + 4 (𝜕𝑦)2 .

𝜕𝑧 𝜕𝑧
Thus, (𝜕𝑥)2 + (𝜕𝑦)2 = 4𝑧 𝑜𝑟 𝑝2 + 𝑞 2 = 4𝑧.

2.Form the PDE from the following by eliminating the constants

𝑧 = (𝑥 2 + 𝑎)(𝑦 2 + 𝑏). 𝐴𝑛𝑠𝑤𝑒𝑟 𝑝𝑞 = 4𝑥𝑦𝑧.

Hint:

𝜕𝑧 𝜕𝑧
= 2𝑥(𝑦 2 + 𝑏); = 2𝑦(𝑥 2 + 𝑎)
𝜕𝑥 𝜕𝑦

3. Show that the family of spheres 𝑥 2 + 𝑦 2 + (𝑧 − 𝑐)2 = 𝑟 2 where c and r are constants satisfies
the first order partial differential equation 𝑦𝑝 − 𝑥𝑞 = 0 by deriving the associated PDE.

Solution

Differentiating implicitly and partially with respect to x and y, we have

𝜕𝑧
2𝑥 + 2(𝑧 − 𝑐) = 0 𝑜𝑟 2𝑥 + 2(𝑧 − 𝑐)𝑝 = 0 … (𝑖)
𝜕𝑥
𝜕𝑧
2𝑦 + 2(𝑧 − 𝑐) = 0 𝑜𝑟 2𝑦 + 2(𝑧 − 𝑐)𝑞 = 0 … (𝑖𝑖)
𝜕𝑦

Multiplying (i) by q and (ii) by p and eliminating the term with arbitrary constant, we have

2𝑦𝑝 + 2(𝑧 − 𝑐)𝑝𝑞 = 0

2𝑥𝑞 + 2(𝑧 − 𝑐)𝑝𝑞 = 0

2𝑦𝑝 − 2𝑥𝑞 = 0

Thus, 𝑦𝑝 − 𝑥𝑞 = 0.

4.Find the differential equation of the family of spheres of radius 5 with centres on the plane 𝑥 =
𝑦, whose equation is (𝑥 − 𝑎)2 + (𝑦 − 𝑎)2 + (𝑧 − 𝑏)2 = 25, 𝑎 and b being arbitrary constants.

𝐴𝑛𝑠𝑤𝑒𝑟 (𝑥 − 𝑦)2 (𝑝2 + 𝑞 2 + 1) = 25(𝑝 − 𝑞)2 .

11
Solution

Differentiating implicitly and partially with respect to x and y, we have

𝜕𝑧
2(𝑥 − 𝑎) + 2(𝑧 − 𝑏) = 0 𝑜𝑟 (𝑥 − 𝑎) + (𝑧 − 𝑏)𝑝 = 0; 𝑎𝑛𝑑
𝜕𝑥
𝜕𝑧
2(𝑦 − 𝑎) + 2(𝑧 − 𝑏) = 0 𝑜𝑟 (𝑦 − 𝑎) + (𝑧 − 𝑏)𝑞 = 0.
𝜕𝑦

Let 𝑧 − 𝑏 = −𝑚; 𝑡ℎ𝑒𝑛 𝑥 − 𝑎 = 𝑝𝑚 𝑎𝑛𝑑 𝑦 − 𝑎 = 𝑞𝑚 … (∗)

Making these replacements in the given equation gives (𝑝𝑚)2 + (𝑞𝑚)2 + (−𝑚)2 = 25.

i.e. 𝑚2 (𝑝2 + 𝑞 2 + 1) = 25.

𝑥−𝑦
From equation (*), 𝑥 − 𝑦 = (𝑝 − 𝑞)𝑚, 𝑔𝑖𝑣𝑖𝑛𝑔 𝑚 = 𝑝−𝑞.

(𝑥−𝑦)2
Substitution in the above equation yields (𝑝−𝑞)2 (𝑝2 + 𝑞 2 + 1) = 25, which gives the required
differential equation.

Deriving a Partial differential equation by elimination of arbitrary functions

Partial differential equations may be obtained by elimination of arbitrary functions.

Examples

1.Eliminate the arbitrary functions 𝑓(𝑥)𝑎𝑛𝑑 𝑔(𝑦)𝑓𝑟𝑜𝑚 𝑧 = 𝑦𝑓(𝑥) + 𝑥𝑔(𝑦).

Differentiating partially with respect to x and y gives

𝑝 = 𝑦𝑓 ′ (𝑥) + 𝑔(𝑦) … (1)

𝑞 = 𝑓(𝑥) + 𝑥𝑔′ (𝑦) … (2)

Since it is not possible to eliminate 𝑓, 𝑔, 𝑓 ′ 𝑎𝑛𝑑 𝑔′ from (1) and (2) and the given equation, we
find the second partial derivatives.

Differentiating (1) with respect to x, 𝑟 = 𝑦𝑓 ′′ (𝑥) … (3).

Differentiating (1) with respect to y, 𝑠 = 𝑓 ′ (𝑥) + 𝑔′ (𝑦) … (4).

Differentiating (2) with respect to y, 𝑡 = 𝑥𝑔′′ (𝑦) … (5).

1 1
From (1) and (2), we find 𝑓 ′ (𝑥) = 𝑦 [𝑝 − 𝑔(𝑦)] 𝑎𝑛𝑑 𝑔′ (𝑦) = 𝑥 [𝑞 − 𝑓(𝑥)].

12
 1 1
Hence from equation (4), 𝑠 = 𝑓 ′ (𝑥) + 𝑔′ (𝑦) = 𝑦 [𝑝 − 𝑔(𝑦)] + 𝑥 [𝑞 − 𝑓(𝑥)].

Thus, 𝑥𝑦𝑠 = 𝑥[𝑝 − 𝑔(𝑦)] + 𝑦[𝑞 − 𝑓(𝑥)] 𝑜𝑟

𝑥𝑦𝑠 = 𝑝𝑥 + 𝑞𝑦 − [𝑦𝑓(𝑥) + 𝑥𝑔(𝑦)] = 𝑝𝑥 + 𝑞𝑦 − 𝑧.

𝜕2𝑧 𝜕𝑧 𝜕𝑧
Therefore 𝑥𝑦𝑠 = 𝑝𝑥 + 𝑞𝑦 − 𝑧 𝑜𝑟 𝑥𝑦 𝜕𝑥𝜕𝑦 = 𝑥 𝜕𝑥 + 𝑦 𝜕𝑦 − 𝑧 is the resulting partial differential
equation.

2.Eliminate the arbitrary function ∅(𝑥 + 𝑦)𝑓𝑟𝑜𝑚 𝑧 = ∅(𝑥 + 𝑦).

Let 𝑥 + 𝑦 = 𝑢 so that the given relation is 𝑧 = ∅(𝑢).

Differentiating with respect to x and y yields

𝑑∅ 𝜕𝑢 𝑑∅ 𝜕𝑢
𝑝= = ∅′ (𝑢)𝑎𝑛𝑑 𝑞 = = ∅′ (𝑢)
𝑑𝑢 𝜕𝑥 𝑑𝑢 𝜕𝑦
𝜕𝑧 𝜕𝑧
Thus 𝑝 = 𝑞 𝑜𝑟 = 𝜕𝑦 is the resulting partial differential equation.
𝜕𝑥

Exercise

1.Eliminate the arbitrary function in the following and hence obtain the corresponding PDE.

𝑧 = 𝑥 + 𝑦 + 𝑓(𝑥𝑦).

Solution

Differentiating partially with respect to x and y, we have

𝜕𝑧
= 1 + 𝑦𝑓 ′ (𝑥𝑦) … (𝑖)
𝜕𝑥
𝜕𝑧
= 1 + 𝑥𝑓 ′ (𝑥𝑦) … (𝑖𝑖)
𝜕𝑦

Multiplying equation (i) by x and (ii) by y gives

𝜕𝑧
𝑥 = 𝑥 + 𝑥𝑦𝑓 ′ (𝑥𝑦)
𝜕𝑥
𝜕𝑧
𝑦 = 𝑦 + 𝑥𝑦𝑓 ′ (𝑥𝑦)
𝜕𝑦
𝜕𝑧 𝜕𝑧
Subtraction yields 𝑥 𝜕𝑥 − 𝑦 𝜕𝑦 = 𝑥 − 𝑦, the required partial differential equation.

13
2.All surfaces of revolution with Z – axis as the axis of symmetry satisfy the equation

𝑧 = 𝑓(𝑥 2 + 𝑦 2 ), 𝑤ℎ𝑒𝑟𝑒 𝑓 is arbitrary function. Obtain a partial differential equation

corresponding to this relation by eliminating the arbitrary function.

Solution

Differentiating partially with respect to x and y, we have

𝜕𝑧
= 2𝑥𝑓 ′ (𝑥 2 + 𝑦 2 ) … (𝑖)
𝜕𝑥
𝜕𝑧
= 2𝑦𝑓 ′ (𝑥 2 + 𝑦 2 ) … (𝑖𝑖)
𝜕𝑦

Multiplying equation (i) by y and (ii) by x gives

𝜕𝑧
𝑦 = 2𝑥𝑦𝑓 ′ (𝑥 2 + 𝑦 2 )
𝜕𝑥
𝜕𝑧
𝑥 = 2𝑥𝑦𝑓 ′ (𝑥 2 + 𝑦 2 )
𝜕𝑦
𝜕𝑧 𝜕𝑧
Subtracting the two equations gives 𝑦 𝜕𝑥 − 𝑥 𝜕𝑦 = 0, which is the required partial differential
equation.

3.Eliminate the arbitrary function from the following and hence obtain the corresponding partial
differential equation.

𝑧 = 𝑥𝑦 + 𝑓(𝑥 2 + 𝑦 2 ) 𝐴𝑛𝑠𝑤𝑒𝑟 𝑦𝑝 − 𝑥𝑞 = 𝑦 2 − 𝑥 2

General solution of first order partial differential equations

First order partial differential equations, for which the general solution can be obtained

directly, can be divided in to the following categories.

(1) Equations that can be solved by direct (partial) integration.

(2) Lagrange’s linear equation of the first order.

Equations that can be solved by direct (partial)integration

𝜕𝑧
1.Solve the partial differential equation 𝜕𝑥 = 2𝑥 + 3𝑦.

14
Integrating partially with respect to x, we have 𝑧 = 𝑥 2 + 3𝑥𝑦 + ∅(𝑦), 𝑤ℎ𝑒𝑟𝑒 ∅ is an arbitrary
function.
𝜕𝑧 𝜕𝑧
(2)Solve the equations 𝜕𝑥 = 3𝑥 − 𝑦 and 𝜕𝑦 = −𝑥 + cos 𝑦 simultaneously.

Given

𝜕𝑧
= 3𝑥 − 𝑦 … (𝑎)
𝜕𝑥
𝜕𝑧
= −𝑥 + cos 𝑦 … (𝑏)
𝜕𝑦

Integrating (a) partially with respect to x,

3𝑥 2
𝑧= − 𝑦𝑥 + 𝑓(𝑦) … (𝑐)
2
Differentiating (c) partially with respect to y, we have

𝜕𝑧
= −𝑥 + 𝑓 ′ (𝑦) … (𝑑)
𝜕𝑦

Comparing (b) and (d), we get

𝑓 ′ (𝑦) = cos 𝑦 𝑖. 𝑒. 𝑓(𝑦) = sin 𝑦 + 𝑐

Therefore the required solution is

3𝑥 2
𝑧= − 𝑦𝑥 + sin 𝑦 + 𝑐 , 𝑤ℎ𝑒𝑟𝑒 𝑐 𝑖𝑠 𝑎𝑛 𝑎𝑟𝑏𝑖𝑡𝑟𝑎𝑟𝑦 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡.
2
Lagrange’s linear equation of the first order

A linear partial differential equation of the first order , which is of the form 𝑃𝑝 + 𝑄𝑞 = 𝑅 … (∗)

where P, Q, R are functions of 𝑥, 𝑦, 𝑧 is called Lagrange’s linear equation.

Lagrange reduced the problem of finding the general solution of (*) to that of solving an
auxiliary system (called the Lagrange system) of ordinary differential equations. His method of
solution is summarized in three steps as highlighted below.

(1)To solve 𝑃𝑝 + 𝑄𝑞 = 𝑅, we form the corresponding subsidiary simultaneous equations

𝑑𝑥 𝑑𝑦 𝑑𝑧
= = .
𝑃 𝑄 𝑅

15
(2)Solving these equations, we get two independent solutions 𝑢(𝑥, 𝑦, 𝑧) = 𝑎 𝑎𝑛𝑑 𝑣(𝑥, 𝑦, 𝑧) = 𝑏.

(3)Then the required general solution is 𝑓(𝑢, 𝑣) = 0 𝑜𝑟 𝑢 = ∅(𝑣)𝑜𝑟 𝑣 = 𝜑(𝑢).

𝑑𝑥 𝑑𝑦 𝑑𝑧
Solution of the simultaneous equations = = .
𝑃 𝑄 𝑅

Methods of grouping:

By grouping any two of three ratios, it may be possible to get an ordinary differential equation
containing only two variables, even though P;Q;R are in general, functions of x,y,z. By solving
this equation, we can get a solution of the simultaneous equations. By this method, we may be
able to get two independent solutions, by using different groupings.

Methods of multipliers:

If we can find a set of three quantities l,m,n which may be constants or functions of the variables
x,y,z, such that 𝑙𝑃 + 𝑚𝑄 + 𝑛𝑅 = 0, then the solution of the simultaneous equation is found out
as follows.

𝑑𝑥 𝑑𝑦 𝑑𝑧 𝑙𝑑𝑥 + 𝑚𝑑𝑦 + 𝑛𝑑𝑧

= = =
𝑃 𝑄 𝑅 𝑙𝑃 + 𝑚𝑄 + 𝑛𝑅

Since 𝑙𝑃 + 𝑚𝑄 + 𝑛𝑅 = 0, 𝑙𝑑𝑥 + 𝑚𝑑𝑦 + 𝑛𝑑𝑧 = 0. If 𝑙𝑑𝑥 + 𝑚𝑑𝑦 + 𝑛𝑑𝑧 = 0 is an exact

differential of some function 𝑢(𝑥, 𝑦, 𝑧) = 0, then we get 𝑑𝑢 = 0. Integrating this, we get

𝑑𝑥 𝑑𝑦 𝑑𝑧
𝑢 = 𝑎, which is a solution of = = .
𝑃 𝑄 𝑅

Similarly, if we can find another set of independent multipliers 𝑙 ′ , 𝑚′ , 𝑛′ , we can get

another independent solution 𝑣 = 𝑏.

Examples

1(a) Solve 𝑥𝑝 + 𝑦𝑞 = 𝑥. .

We are given 𝑥𝑝 + 𝑦𝑞 = 𝑥.

This is Lagrange’s type of PDE where P = x , Q = y , R = x .

𝑑𝑥 𝑑𝑦 𝑑𝑧
The subsidiary equations are = = .
𝑥 𝑦 𝑥

𝑑𝑥 𝑑𝑦
Taking first two members, =
𝑥 𝑦

16
Integrating we get ln 𝑥 = ln 𝑦 + ln 𝑐1
𝑥
i.e. = 𝑐1 ………….(1)
𝑦

𝑥
𝑢=
𝑦
𝑑𝑥 𝑑𝑧
Taking the first and last member, = .
𝑥 𝑥

i.e. 𝑑𝑥 = 𝑑𝑧.

Integrating we get 𝑥 − 𝑧 = 𝑐2 ………......(2)

𝑣 =𝑥−𝑧
𝑥
Therefore the solution of the given PDE is ∅(𝑢, 𝑣) = 0 𝑖. 𝑒. ∅ (𝑦 , 𝑥 − 𝑧) = 0.

e.g. ∅(𝑢, 𝑣) = 𝑢2 + 𝑣 = 0 𝑜𝑟 sin 𝑢 + 𝑣 = 0 are examples of members of this family of arbitrary

functions both of which must satisfy the given partial differential equation.
𝑥 𝑥2
Checking for 𝑢2 + 𝑣 = 0, we have (𝑦)2 + (𝑥 − 𝑧) = 0 0𝑟 𝑧 = 𝑥 + 𝑦 2.

𝜕𝑧 2𝑥 𝜕𝑧 −2𝑥 2
Therefore =1+ , = −2𝑥 2 𝑦 −3 = .
𝜕𝑥 𝑦2 𝜕𝑦 𝑦3

2𝑥 −2𝑥 2 2𝑥 2 −2𝑥 2 𝑦
Hence, 𝑥𝑝 + 𝑦𝑞 = 𝑥 (1 + 𝑦 2 ) + 𝑦 ( ) = (𝑥 + )+( ) = 𝑥, as expected.
𝑦3 𝑦2 𝑦3

𝑥 𝑥
Checking for sin 𝑢 + 𝑣 = 0, we have sin( 𝑦) + (𝑥 − 𝑧) = 0 0𝑟 𝑧 = 𝑥 + sin( 𝑦).

𝜕𝑧 1 𝑥 𝜕𝑧 −𝑥 𝑥
Therefore 𝜕𝑥 = 1 + y cos(𝑦) , 𝜕𝑦 = cos(𝑦).
𝑦2

1 𝑥 −𝑥 𝑥 x 𝑥 −𝑥𝑦 𝑥
Hence, 𝑥𝑝 + 𝑦𝑞 = 𝑥 (1 + y cos(𝑦)) + 𝑦 ( 𝑦 2 cos(𝑦)) = (𝑥 + y cos(𝑦)) + ( 𝑦 2 cos(𝑦)) = 𝑥, as
expected.
𝑥
Thus, the general solution to the given partial differential equation is ∅ (𝑦 , 𝑥 − 𝑧) = 0.

1(b) Find the general solution of 𝑝𝑥 + 𝑞𝑦 = 3𝑧.

We are given 𝑝𝑥 + 𝑞𝑦 = 3𝑧.

This is Lagrange’s type of PDE where P = x , Q = y , R = 3z .

𝑑𝑥 𝑑𝑦 𝑑𝑧
The auxiliary system is = = .
𝑥 𝑦 3𝑧

17
 𝑑𝑥 𝑑𝑦
From = , ln 𝑥 = ln 𝑦 + 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑜𝑟 ln 𝑥 − ln 𝑦 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡.
𝑥 𝑦

𝑥 𝑥
Hence we obtain ln( 𝑦) = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑜𝑟 = 𝑎.
𝑦

𝑥
Thus, 𝑢 = 𝑦.

𝑑𝑥 𝑑𝑧 1
From = 3𝑧 , ln 𝑥 = 3 ln 𝑧 + 𝑘 𝑜𝑟 3ln 𝑥 − ln 𝑧 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑖. 𝑒. ln( 𝑥 3 ) − ln 𝑧 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡.
𝑥

𝑥3 𝑥3
Hence we obtain ln( 𝑧 ) = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑜𝑟 = 𝑏.
𝑧

𝑥3
Thus, 𝑣 = .
𝑧

𝑥 𝑥3
Therefore the general solution of the given PDE is ∅(𝑢, 𝑣) = 0 𝑖. 𝑒. ∅ (𝑦 , 𝑧 ) = 0.

𝑑𝑦 𝑑𝑧 𝑦3 𝑥 𝑦3 𝑥3 𝑦3
Notice that from = 3𝑧 we obtain = 𝑐 and we may write 𝜑 (𝑦 , ) = 0 𝑜𝑟 𝛽 ( 𝑧 , ) = 0,
𝑦 𝑧 𝑧 𝑧

where 𝜑 𝑎𝑛𝑑 𝛽 are arbitrary functions. However, ∅, 𝜑 𝑎𝑛𝑑 𝛽 are all equivalent and we shall call
any one of them the general solution.

1(c) Find the general solution of 2𝑝 + 3𝑞 = 1.

This is Lagrange’s type of PDE where P = 2 , Q = 3 , R = 1 .

𝑑𝑥 𝑑𝑦 𝑑𝑧
The auxiliary system is = = .
2 3 1

𝑑𝑥 𝑑𝑦 𝑥 𝑦
From = , we have on integration, 2 = 3 + 𝑘 𝑜𝑟 3𝑥 − 2𝑦 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡.
2 3

Thus, 3𝑥 − 2𝑦 = 𝑎, 𝑖. 𝑒. 𝑢 = 3𝑥 − 2𝑦.
𝑑𝑥 𝑑𝑧 𝑥
From = , we have on integration, 2 = 𝑧 + 𝑘 ′ 𝑜𝑟 𝑥 − 2𝑧 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡.
2 1

Thus, 𝑥 − 2𝑧 = 𝑏, 𝑖. 𝑒. 𝑣 = 𝑥 − 2𝑧.

Hence the general solution of the given PDE is ∅(3𝑥 − 2𝑦, 𝑥 − 2𝑧) = 0.

Exercise

Use the functions given in 1(a) to check the solution in 1(c) above.
𝜕𝑧 𝜕𝑧
1(d) Find the general solution of 𝑦 2 𝑧 𝜕𝑥 − 𝑥 2 𝑧 𝜕𝑦 = 𝑥 2 𝑦.

We are given 𝑦 2 𝑧𝑝 − 𝑥 2 𝑧𝑞 = 𝑥 2 𝑦.

18
This is Lagrange’s type of PDE where P = 𝑦 2 𝑧 , Q = −𝑥 2 𝑧, R = 𝑥 2 𝑦
𝑑𝑥 𝑑𝑦 𝑑𝑧
The subsidiary equations are 𝑦 2𝑧 = −𝑥 2 𝑧 = 𝑥 2 𝑦.

𝑑𝑥 𝑑𝑦 𝑑𝑥 𝑑𝑦
From 𝑦 2𝑧 = −𝑥 2 𝑧, we get 𝑦 2 = −𝑥 2 𝑜𝑟 𝑥 2 𝑑𝑥 + 𝑦 2 𝑑𝑦 = 0 which on integration gives

𝑥3 𝑦3
+ = 𝑘 𝑜𝑟 𝑥 3 + 𝑦 3 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡.
3 3

i.e.𝑥 3 + 𝑦 3 = 𝑎 𝑜𝑟 𝑢 = 𝑥 3 + 𝑦 3 .
𝑑𝑦 𝑑𝑧 𝑑𝑦 𝑑𝑧
From = , we get = 𝑜𝑟 𝑦𝑑𝑦 + 𝑧𝑑𝑧 = 0 which on integration gives
−𝑥 2 𝑧 𝑥2𝑦 −𝑧 𝑦

𝑦2 𝑧2
+ = 𝑘 ′ 𝑜𝑟 𝑦 2 + 𝑧 2 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡.
2 2

i.e.𝑦 2 + 𝑧 2 = 𝑏 𝑜𝑟 𝑣 = 𝑦 2 + 𝑧 2 .

Thus, the general solution is ∅(𝑢, 𝑣) = ∅(𝑥 3 + 𝑦 3 , 𝑦 2 + 𝑧 2 ) = 0.

𝜕𝑧 𝜕𝑧
1(e) Solve the partial differential equation 𝑧 𝜕𝑥 − 𝑧 𝜕𝑦 = 𝑧 2 + (𝑥 + 𝑦)2.

We are given 𝑧𝑝 − 𝑧𝑞 = 𝑧 2 + (𝑥 + 𝑦)2 .

This is Lagrange’s type of PDE where P = 𝑧 , Q = −𝑧, R = 𝑧 2 + (𝑥 + 𝑦)2 .

𝑑𝑥 𝑑𝑦 𝑑𝑧
The subsidiary equations are = −𝑧 = .
𝑧 𝑧 2 +(𝑥+𝑦)2

𝑑𝑥 𝑑𝑦
Taking = −𝑧 , 𝑤𝑒 𝑔𝑒𝑡 𝑑𝑥 + 𝑑𝑦 = 0 𝑜𝑟 𝑥 + 𝑦 = 𝑐1 , 𝑎 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡, 𝑢𝑝𝑜𝑛 𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑖𝑜𝑛.
𝑧

Thus, 𝑢 = 𝑥 + 𝑦.
𝑑𝑥 𝑑𝑧 𝑑𝑥 𝑑𝑧
Taking = 𝑧 2+(𝑥+𝑦)2 , and using the above equation 𝑥 + 𝑦 = 𝑐1 , we can write = 𝑧 2+(𝑐 2
.
𝑧 𝑧 1)

𝑧𝑑𝑧 2𝑧𝑑𝑧
i.e. 𝑑𝑥 = 𝑜𝑟 2𝑑𝑥 = , which on integration gives 2𝑥 = ln(𝑧 2 + 𝑐1 2 ) + 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡
𝑧 2 +(𝑐1 )2 𝑧 2 +(𝑐1 )2

i.e. ln(𝑧 2 + 𝑐1 2 ) − 2𝑥 = 𝑐2 𝑜𝑟 𝑣 = ln(𝑧 2 + 𝑐1 2 ) − 2𝑥 = 𝑙𝑛[𝑧 2 + (𝑥 + 𝑦)2 ] − 2𝑥.

Hence the general solution is ∅(𝑢, 𝑣) = ∅(𝑥 + 𝑦, 𝑙𝑛[𝑧 2 + (𝑥 + 𝑦)2 ] − 2𝑥) = 0.

𝑑𝑥 𝑑𝑦 𝑑𝑧
Note that by grouping any two of three ratios of the auxiliary system = = , we get an
𝑃 𝑄 𝑅
easily solved ordinary differential equation containing only two variables, even though P;Q;R are
in general, functions of x,y,z.

19
Note also that this method of regrouping can be extended readily to solve linear first order
differential equations involving more than two independent variables.
𝜕𝑧 𝜕𝑧 𝜕𝑧
1(f) Find the general solution of 𝑥 𝜕𝑥 + 𝑦 𝜕𝑦 + 𝑡 𝜕𝑡 = 𝑥𝑦𝑡, 𝑧 𝑏𝑒𝑖𝑛𝑔 𝑡ℎ𝑒 𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒.

𝑑𝑥 𝑑𝑦 𝑑𝑡 𝑑𝑧
The auxiliary system is = = = 𝑥𝑦𝑡 … (∗).
𝑥 𝑦 𝑡

𝑑𝑥 𝑑𝑦 𝑥
From = , ln 𝑥 = ln 𝑦 + 𝑘 𝑜𝑟 ln 𝑥 − ln 𝑦 = 𝑘, 𝑖. 𝑒. ln 𝑦 = 𝑘.
𝑥 𝑦

𝑥 𝑥
Therefore, 𝑦 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 = 𝑎 𝑜𝑟 𝑢 = 𝑦.

𝑑𝑦 𝑑𝑡 𝑦
From = , ln 𝑦 = ln 𝑡 + 𝑘′ 𝑜𝑟 ln 𝑦 − ln 𝑡 = 𝑘′, 𝑖. 𝑒. ln 𝑡 = 𝑘′.
𝑦 𝑡

𝑦 𝑦
Therefore, = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 = 𝑏 𝑜𝑟 𝑣 = 𝑡 .
𝑡

A third independent solution may be obtained using multipliers 𝑦𝑡, 𝑥𝑡, 𝑥𝑦, −3.
𝑦𝑡𝑑𝑥+𝑥𝑡𝑑𝑦+𝑥𝑦𝑑𝑡−3𝑑𝑧
Therefore, each ratio in equation (*) is equal to 0

Thus, 𝑦𝑡𝑑𝑥 + 𝑥𝑡𝑑𝑦 + 𝑥𝑦𝑑𝑡 − 3𝑑𝑧 = 0 𝑜𝑟 𝑑(𝑥𝑦𝑡) − 3𝑑𝑧 = 𝑜.

𝑑(𝑥𝑦𝑡) = 3𝑑𝑧 𝑜𝑟 𝑥𝑦𝑡 = 3𝑧 + 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡.

Therefore, 𝑥𝑦𝑡 − 3𝑧 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 = 𝑐 𝑜𝑟 𝑤 = 𝑥𝑦𝑡 − 3𝑧.

𝑥 𝑦
Hence the general solution is ∅(𝑢, 𝑣, 𝑤) = 0 𝑜𝑟 ∅ (𝑦 , 𝑡 , 𝑥𝑦𝑡 − 3𝑧) = 0.

2(a) Solve the equation (𝑥 − 2𝑧)𝑝 + (2𝑧 − 𝑦)𝑞 = 𝑦 − 𝑥

We are given(𝑥 − 2𝑧)𝑝 + (2𝑧 − 𝑦)𝑞 = 𝑦 − 𝑥.

This is Lagrange’s type of PDE where 𝑃 = 𝑥 − 2𝑧, 𝑄 = 2𝑧 − 𝑦, 𝑅 = 𝑦 − 𝑥. .

𝑑𝑥 𝑑𝑦 𝑑𝑧
The subsidiary equations are = = . ……….(1)
𝑥−2𝑧 2𝑧−𝑦 𝑦−𝑥

𝑑𝑥+𝑑𝑦+𝑑𝑧
Using the multipliers 1,1,1, each ratio in (1) = .
0

i.e. 𝑑𝑥 + 𝑑𝑦 + 𝑑𝑧 = 0.

Integrating, we get 𝑥 + 𝑦 + 𝑧 = 𝑎, 𝑖. 𝑒. 𝑢 = 𝑥 + 𝑦 + 𝑧.
𝑦𝑑𝑥+𝑥𝑑𝑦+2𝑧𝑑𝑧
Using the multipliers y,x,2z, each ratio in (1) = .
0

20
i.e. 𝑑(𝑥𝑦) + 2𝑧𝑑𝑧 = 0

Integrating, we get 𝑥𝑦 + 𝑧 2 = 𝑏, 𝑖. 𝑒. 𝑣 = 𝑥𝑦 + 𝑧 2 .

Therefore the general solution of the given partial differential equation is

𝑓(𝑥 + 𝑦 + 𝑧, 𝑥𝑦 + 𝑧 2 ) = 0.

2(b) Find the general solution of (𝑦 − 𝑧)𝑝 + (𝑥 − 𝑦)𝑞 = 𝑧 − 𝑥.

This is Lagrange’s type of PDE where 𝑃 = 𝑦 − 𝑧, 𝑄 = 𝑥 − 𝑦, 𝑅 = 𝑧 − 𝑥. .

𝑑𝑥 𝑑𝑦 𝑑𝑧
The auxiliary system is = = .
𝑦−𝑧 𝑥−𝑦 𝑧−𝑥

𝑑𝑥+𝑑𝑦+𝑑𝑧
Using the multipliers 1,1,1, each ratio in the auxiliary system is equal to .
0

i.e. 𝑑𝑥 + 𝑑𝑦 + 𝑑𝑧 = 0, 𝑔𝑖𝑣𝑖𝑛𝑔 𝑥 + 𝑦 + 𝑧 = 𝑎, upon integration.

Therefore, 𝑢 = 𝑥 + 𝑦 + 𝑧.

We choose the multipliers x,z,y, since 𝑥𝑃 + 𝑧𝑄 + 𝑦𝑅 = 𝑥(𝑦 − 𝑧) + 𝑧(𝑥 − 𝑦) + 𝑦(𝑧 − 𝑥) = 0.

𝑥𝑑𝑥+𝑧𝑑𝑦+𝑦𝑑𝑧
Again using the multipliers x,z,y, each ratio in the auxiliary system is equal to .
0

Therefore, 𝑥𝑑𝑥 + 𝑧𝑑𝑦 + 𝑦𝑑𝑧 = 0 𝑜𝑟 𝑥𝑑𝑥 + 𝑑(𝑦𝑧) = 0.

𝑥2
Integrating, + 𝑦𝑧 = 𝑘 𝑜𝑟 𝑥 2 + 2𝑦𝑧 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡.
2

Thus, 𝑣 = 𝑥 2 + 2𝑦𝑧, and hence the general solution is ∅(𝑥 + 𝑦 + 𝑧, 𝑥 2 + 2𝑦𝑧) = 0.

2(c) Find the general solution of (𝑥 2 − 𝑦 2 − 𝑧 2 )𝑝 + 2𝑥𝑦𝑞 = 2𝑥𝑧.

This is Lagrange’s type of PDE where 𝑃 = 𝑥 2 − 𝑦 2 − 𝑧 2 , 𝑄 = 2𝑥𝑦, 𝑅 = 2𝑥𝑧.

𝑑𝑥 𝑑𝑦 𝑑𝑧
The auxiliary system is 𝑥 2 −𝑦 2−𝑧 2 = 2𝑥𝑦 = 2𝑥𝑧.

𝑑𝑦 𝑑𝑧 𝑑𝑦 𝑑𝑧
From 2𝑥𝑦 = 2𝑥𝑧 , 𝑤𝑒 𝑔𝑒𝑡 = 𝑓𝑟𝑜𝑚 𝑤ℎ𝑖𝑐ℎ 𝑤𝑒 𝑜𝑏𝑡𝑎𝑖𝑛
𝑦 𝑧

𝑦
ln 𝑦 = ln 𝑧 + 𝑘 𝑜𝑟 ln 𝑦 − ln 𝑧 = ln ( ) = 𝑘.
𝑧
𝑦 𝑦
i.e 𝑧 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑜𝑟 𝑢 = 𝑧 .

We choose the multipliers x,y,z, since 𝑥𝑃 + 𝑦𝑄 + 𝑧𝑅 = 𝑥(𝑥 2 − 𝑦 2 − 𝑧 2 ) + 2𝑥𝑦 2 + 2𝑥𝑧 2

21
𝑥 3 − 𝑥𝑦 2 − 𝑥𝑧 2 + 2𝑥𝑦 2 + 2𝑥𝑧 2 = 𝑥 3 + 𝑥𝑦 2 + 𝑥𝑧 2 = 𝑥(𝑥 2 + 𝑦 2 + 𝑧 2 ).
𝑥𝑑𝑥+𝑦𝑑𝑦+𝑧𝑑𝑧
Thus, each ratio in the auxiliary system is equal to .
𝑥(𝑥 2 +𝑦 2 +𝑧 2 )

𝑑𝑧 𝑥𝑑𝑥+𝑦𝑑𝑦+𝑧𝑑𝑧 𝑑𝑧 2𝑥𝑑𝑥+2𝑦𝑑𝑦+2𝑧𝑑𝑧 𝑑(𝑥 2 +𝑦 2 +𝑧 2 )

Taking 2𝑥𝑧 = , 𝑤𝑒 𝑜𝑏𝑡𝑎𝑖𝑛 = =
𝑥(𝑥 2 +𝑦 2 +𝑧 2 ) 𝑧 (𝑥 2 +𝑦 2 +𝑧 2 ) (𝑥 2 +𝑦 2 +𝑧 2 )

which on integration yields ln 𝑧 = ln(𝑥 2 + 𝑦 2 + 𝑧 2 ) + 𝑘 𝑜𝑟 ln(𝑥 2 + 𝑦 2 + 𝑧 2 ) − ln 𝑧 = 𝑘′

𝑥 2 +𝑦 2 +𝑧 2 𝑥 2 +𝑦 2 +𝑧 2 𝑥 2 +𝑦 2 +𝑧 2
i.e. ln( ) = 𝑘′ 𝑜𝑟 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡, and thus 𝑣 = .
𝑧 𝑧 𝑧

𝑦 𝑥 2 +𝑦 2 +𝑧 2
Hence the general solution is ∅ ( , ) = 0.
𝑧 𝑧

2(d) Find the general integral of the following linear partial differential equations

(i)𝑦 2 𝑝 − 𝑥𝑦𝑞 = 𝑥(𝑧 − 2𝑦)

The auxiliary equations are

𝑑𝑥 𝑑𝑦 𝑑𝑧
2
= =
𝑦 −𝑥𝑦 𝑥(𝑧 − 2𝑦)
𝑑𝑥 𝑑𝑦
The first two members give = −𝑥 𝑜𝑟 𝑥𝑑𝑥 = −𝑦𝑑𝑦.
𝑦

𝑥2 𝑦2
Therefore = + 𝑐 𝑜𝑟 𝑥 2 + 𝑦 2 = 𝑐1.
2 2

𝑢 = 𝑥2 + 𝑦2.
𝑑𝑦 𝑑𝑧
The last two members give = 𝑧−2𝑦 𝑜𝑟 𝑧𝑑𝑦 − 2𝑦𝑑𝑦 = −𝑦𝑑𝑧. Thus
−𝑦

2𝑦𝑑𝑦 = 𝑦𝑑𝑧 + 𝑧𝑑𝑦 𝑜𝑟 2𝑦𝑑𝑦 = 𝑑(𝑦𝑧) 𝑤ℎ𝑖𝑐ℎ 𝑜𝑛 𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑖𝑜𝑛 𝑦𝑖𝑒𝑙𝑑𝑠

𝑦 2 = 𝑦𝑧 + 𝑐2 𝑜𝑟 𝑦 2 − 𝑦𝑧 = 𝑐2 𝑖. 𝑒. 𝑣 = 𝑦 2 − 𝑦𝑧.

Hence ∅(𝑥 2 + 𝑦 2 , 𝑦 2 − 𝑦𝑧) = 0 is the required solution.

(ii)(𝑦 + 𝑧𝑥)𝑝 − (𝑥 + 𝑦𝑧)𝑞 = 𝑥 2 − 𝑦 2

The auxiliary equations are

𝑑𝑥 𝑑𝑦 𝑑𝑧
= = 2
𝑦 + 𝑧𝑥 −(𝑥 + 𝑦𝑧) 𝑥 − 𝑦 2
𝑥𝑑𝑥+𝑦𝑑𝑦 𝑑𝑧 𝑥𝑑𝑥+𝑦𝑑𝑦 𝑑𝑧
Consider the first combination as 𝑥𝑦+𝑧𝑥 2−𝑥𝑦−𝑦 2𝑧 = 𝑥 2 −𝑦 2 𝑜𝑟 = 𝑥 2 −𝑦 2
𝑧(𝑥 2 −𝑦 2 )

22
 𝑥2 𝑦2 𝑧2
Therefore 𝑥𝑑𝑥 + 𝑦𝑑𝑦 = 𝑧𝑑𝑧 𝑜𝑟 + − = 𝑐 𝑖. 𝑒. 𝑥 2 + 𝑦 2 − 𝑧 2 = 𝑐1
2 2 2

𝑢 = 𝑥2 + 𝑦2 − 𝑧2.

Similarly, consider a second combination as

𝑦𝑑𝑥 + 𝑥𝑑𝑦 𝑑𝑧 𝑦𝑑𝑥 + 𝑥𝑑𝑦 𝑑𝑧 −𝑑𝑧

= 2 𝑜𝑟 = 2 = 2
𝑦2 2
+ 𝑥𝑦𝑧 − 𝑥 − 𝑥𝑦𝑧 𝑥 − 𝑦 2 2
𝑦 −𝑥 2 𝑥 −𝑦 2 𝑦 − 𝑥2

Therefore 𝑦𝑑𝑥 + 𝑥𝑑𝑦 = −𝑑𝑧 𝑜𝑟 𝑑(𝑥𝑦) + 𝑑𝑧 = 0, which on integration gives 𝑥𝑦 + 𝑧 = 𝑐2 .

𝑣 = 𝑥𝑦 + 𝑧

Thus ∅(𝑥 2 + 𝑦 2 − 𝑧 2 , 𝑥𝑦 + 𝑧) = 0 is the required solution.

2(e) Find the general integral of the following linear PDE

(𝑥 2 − 𝑦𝑧)𝑝 + (𝑦 2 − 𝑧𝑥)𝑞 = 𝑧 2 − 𝑥𝑦
𝑑𝑥 𝑑𝑦 𝑑𝑧
The auxiliary system is 𝑥 2 −𝑦𝑧 = 𝑦 2−𝑧𝑥 = 𝑧 2−𝑥𝑦 which is equivalent to

𝑑𝑥 − 𝑑𝑦 𝑑𝑦 − 𝑑𝑧 𝑑𝑧 − 𝑑𝑥
= = 𝑜𝑟
𝑥 2 − 𝑦𝑧 − 𝑦 2 + 𝑧𝑥 𝑦 2 − 𝑧𝑥 − 𝑧 2 + 𝑥𝑦 𝑧 2 − 𝑦𝑥 − 𝑥 2 + 𝑦𝑧

𝑑𝑥 − 𝑑𝑦 𝑑𝑦 − 𝑑𝑧 𝑑𝑧 − 𝑑𝑥
= = 𝑜𝑟
(𝑥 − 𝑦)(𝑥 + 𝑦 + 𝑧) (𝑦 − 𝑧)(𝑥 + 𝑦 + 𝑧) (𝑧 − 𝑥)(𝑥 + 𝑦 + 𝑧)

𝑑𝑥 − 𝑑𝑦 𝑑𝑦 − 𝑑𝑧 𝑑𝑧 − 𝑑𝑥
= =
𝑥−𝑦 𝑦−𝑧 𝑧−𝑥
𝑑(𝑥−𝑦) 𝑑(𝑦−𝑧)
From the first pair, = and integration yields ln(𝑥 − 𝑦) = ln(𝑦 − 𝑧) + ln 𝑐1 𝑜𝑟
𝑥−𝑦 𝑦−𝑧

𝑥−𝑦 𝑥−𝑦
ln(𝑥 − 𝑦) − ln(𝑦 − 𝑧) = ln 𝑐1 𝑜𝑟 ln( ) = ln 𝑐1 𝑖. 𝑒 = 𝑐1
𝑦−𝑧 𝑦−𝑧
𝑥−𝑦
𝑢=
𝑦−𝑧
𝑑(𝑦−𝑧) 𝑑(𝑧−𝑥)
Similarly, from the last pair, = and integration yields
𝑦−𝑧 𝑧−𝑥

ln(𝑦 − 𝑧) = ln(𝑧 − 𝑥) + ln 𝑐2 𝑜𝑟
𝑦−𝑧 𝑦−𝑧
ln(𝑦 − 𝑧) − ln(𝑧 − 𝑥) = ln 𝑐2 𝑜𝑟 ln( ) = ln 𝑐2 𝑖. 𝑒 = 𝑐1
𝑧−𝑥 𝑧−𝑥

23
 𝑦−𝑧
𝑣=
𝑧−𝑥
𝑥−𝑦 𝑦−𝑧
Thus ∅ (𝑦−𝑧 , 𝑧−𝑥 ) = 0 is the required solution.

2(f) Solve (𝑦 + 𝑧)𝑝 + (𝑧 + 𝑥)𝑞 = 𝑥 + 𝑦

The subsidiary or auxiliary equations are

𝑑𝑥 𝑑𝑦 𝑑𝑧
= =
𝑦+𝑧 𝑧+𝑥 𝑥+𝑦
𝑑𝑥+𝑑𝑦+𝑑𝑧 𝑑𝑥−𝑑𝑦 𝑑𝑦−𝑑𝑧
The auxiliary system is equivalent to (𝑦+𝑧)+(𝑧+𝑥)+(𝑥+𝑦) = (𝑦+𝑧)−(𝑧+𝑥) = (𝑧+𝑥)−(𝑥+𝑦) 𝑜𝑟

𝑑𝑥 + 𝑑𝑦 + 𝑑𝑧 𝑑𝑥 − 𝑑𝑦 𝑑𝑦 − 𝑑𝑧
= =
2(𝑥 + 𝑦 + 𝑧) −(𝑥 − 𝑦) −(𝑦 − 𝑧)
𝑑𝑥+𝑑𝑦+𝑑𝑧 𝑑𝑥−𝑑𝑦 𝑑𝑥+𝑑𝑦+𝑑𝑧 𝑑𝑥−𝑑𝑦
From the first pair, = −(𝑥−𝑦) 𝑜𝑟 +2 =0
2(𝑥+𝑦+𝑧) 𝑥+𝑦+𝑧 (𝑥−𝑦)

𝑑(𝑥+𝑦+𝑧) 𝑑(𝑥−𝑦)
Thus +2 = 0 and integration yields ln(𝑥 + 𝑦 + 𝑧) + 2 ln(𝑥 − 𝑦) = ln 𝑐1 𝑜𝑟
𝑥+𝑦+𝑧 (𝑥−𝑦)

ln(𝑥 + 𝑦 + 𝑧) + 𝑙𝑛(𝑥 − 𝑦)2 = ln 𝑐1 𝑜𝑟 ln[(𝑥 + 𝑦 + 𝑧)(𝑥 − 𝑦)2 ] = ln 𝑐1

i.e. (𝑥 + 𝑦 + 𝑧)(𝑥 − 𝑦)2 = 𝑐1 𝑜𝑟 𝑢 = (𝑥 + 𝑦 + 𝑧)(𝑥 − 𝑦)2 .

𝑑𝑥−𝑑𝑦 𝑑𝑦−𝑑𝑧 𝑑𝑥−𝑑𝑦 𝑑𝑦−𝑑𝑧 𝑑(𝑥−𝑦) 𝑑(𝑦−𝑧)
From the last pair, −(𝑥−𝑦) = −(𝑦−𝑧) 𝑜𝑟 (𝑥−𝑦)
= 𝑖. 𝑒. (𝑥−𝑦)
=
(𝑦−𝑧) (𝑦−𝑧)

Integration gives ln(𝑥 − 𝑦) = ln(𝑦 − 𝑧) + ln 𝑐2 𝑜𝑟 ln(𝑥 − 𝑦) − ln(𝑦 − 𝑧) = ln 𝑐2

𝑥−𝑦 𝑥−𝑦 𝑥−𝑦
i.e. 𝑙𝑛 𝑦−𝑧 = ln 𝑐2 𝑜𝑟 = 𝑐2 𝑜𝑟 𝑣 = .
𝑦−𝑧 𝑦−𝑧

𝑥−𝑦
Thus, the general solution is given by ∅ ((𝑥 + 𝑦 + 𝑧)(𝑥 − 𝑦)2 , ) = 0.
𝑦−𝑧

𝜕𝑧 𝜕𝑧
2(g) Find the general solution of the linear equation 𝑥 2 𝜕𝑥 + 𝑦 2 𝜕𝑦 = (𝑥 + 𝑦)𝑧

𝑑𝑥 𝑑𝑦 𝑑𝑧
The subsidiary equations are 𝑥 2 = = 𝑧(𝑥+𝑦)
𝑦2

𝑑𝑥 𝑑𝑦 1 1 1 1 𝑦−𝑥
Taking 𝑥 2 = , integration yields − 𝑥 = − 𝑦 + 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑜𝑟 −𝑦 =𝑎 =
𝑦2 𝑥 𝑥𝑦

24
 𝑦−𝑥
𝑢= .
𝑥𝑦

𝑑𝑥⁄ 𝑑𝑦⁄ 𝑑𝑧⁄

𝑥 𝑦 𝑧
From the auxiliary equations, = =
𝑥 𝑦 𝑥+𝑦

𝑑𝑥⁄ +𝑑𝑦⁄ 𝑑𝑧⁄ 𝑑𝑥 𝑑𝑦 𝑑𝑧

𝑥 𝑦 𝑧
Hence = 𝑜𝑟 + =
𝑥+𝑦 𝑥+𝑦 𝑥 𝑦 𝑧

𝑥𝑦
Integrating, ln 𝑥 + ln 𝑦 = ln 𝑧 + 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑜𝑟 ln(𝑥𝑦) − ln 𝑧 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑜𝑟 ln( 𝑧 ) = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

𝑥𝑦
Therefore, 𝑣 = =𝑏
𝑧

𝑦−𝑥 𝑥𝑦
Hence ∅ ( 𝑥𝑦 , ) = 0 is the general solution.
𝑧

2(h) Solve 𝑝 cos(𝑥 + 𝑦) + 𝑞 sin(𝑥 + 𝑦) = 𝑧

𝑑𝑥 𝑑𝑦 𝑑𝑧
The subsidiary equations are cos(𝑥+𝑦) = sin(𝑥+𝑦) = … (∗)
𝑧

𝑑𝑥+𝑑𝑦 𝑑𝑧 𝑑(𝑥+𝑦) 𝑑𝑧
The auxiliary system is equivalent to cos(𝑥+𝑦)+sin(𝑥+𝑦) = 𝑜𝑟 =
𝑧 cos(𝑥+𝑦)+sin(𝑥+𝑦) 𝑧

Put 𝑥 + 𝑦 = 𝑢 𝑠𝑜 𝑡ℎ𝑎𝑡 𝑑𝑥 + 𝑑𝑦 = 𝑑(𝑥 + 𝑦) = 𝑑𝑢, and thus

𝑑𝑢 𝑑𝑧
= … (1)
cos 𝑢 + sin 𝑢 𝑧
Now,

√2 √2
cos 𝑢 + sin 𝑢 = cos 𝑢. + sin 𝑢.
√2 √2
1 1
= √2 {cos 𝑢. + sin 𝑢. }
√2 √2
𝜋 𝜋
= √2 {cos 𝑢 sin + sin 𝑢 cos }
4 4
But sin(𝐴 + 𝐵) = cos 𝐴 sin 𝐵 + sin 𝐴 cos 𝐵.
𝜋
Therefore cos 𝑢 + sin 𝑢 = √2 {sin(𝑢 + 4 )}. Equation (1) becomes

𝑑𝑢 𝑑𝑧 1 𝜋 𝑑𝑧
𝜋 = 𝑧 𝑜𝑟 𝑐𝑜𝑠𝑒𝑐 (𝑢 + ) 𝑑𝑢 =
4 𝑧
√2 sin(𝑢 + 4) √2

25
 𝜋 𝑑𝑧
Thus ∫ 𝑐𝑜𝑠𝑒𝑐 (𝑢 + 4 ) 𝑑𝑢 = √2 ∫ + 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑖. 𝑒.
𝑧

𝜋 𝜋
𝑙𝑛 [tan (𝑢 + )] = √2 ln 𝑧 + ln 𝑐1 𝑜𝑟𝑙𝑛 [tan (𝑢 + )] − √2 ln 𝑧 = ln 𝑐1 𝑜𝑟
4 4
𝜋
𝜋 tan (𝑢 + 4 )
𝑙𝑛 [tan (𝑢 + )] − ln 𝑧 √2 = ln 𝑐1 𝑜𝑟 ln [ ] = ln 𝑐1 𝑖. 𝑒.
4 𝑧 √2

𝜋 𝜋
tan (𝑢 + 4 ) tan (𝑥 + 𝑦 + 4)
= 𝑐1 𝑜𝑟 = 𝑐1
𝑧 √2 𝑧 √2
Again taking the subsidiary system (∗),

𝑑𝑥 𝑑𝑦 𝑑𝑧
= = 𝑤𝑒 𝑜𝑏𝑡𝑎𝑖𝑛 𝑡ℎ𝑒 𝑒𝑞𝑢𝑖𝑣𝑎𝑙𝑒𝑛𝑡 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛𝑠
cos(𝑥 + 𝑦) sin(𝑥 + 𝑦) 𝑧

𝑑𝑥 + 𝑑𝑦 𝑑𝑥 − 𝑑𝑦 𝑑𝑧
= =
cos(𝑥 + 𝑦) + sin(𝑥 + 𝑦) cos(𝑥 + 𝑦) − sin(𝑥 + 𝑦) 𝑧
𝑑𝑥+𝑑𝑦 𝑑𝑥−𝑑𝑦
From the first pair, cos(𝑥+𝑦)+sin(𝑥+𝑦) = cos(𝑥+𝑦)−sin(𝑥+𝑦) 𝑜𝑟

[cos(𝑥 + 𝑦) − sin(𝑥 + 𝑦)](𝑑𝑥 + 𝑑𝑦)

= 𝑑𝑥 − 𝑑𝑦 𝑖. 𝑒.
cos(𝑥 + 𝑦) + sin(𝑥 + 𝑦)

[cos(𝑥 + 𝑦) − sin(𝑥 + 𝑦)](𝑑𝑥 + 𝑑𝑦)

= 𝑑(𝑥 − 𝑦) … (∗∗)
cos(𝑥 + 𝑦) + sin(𝑥 + 𝑦)
(cos 𝜃−sin 𝜃)𝑑𝜃
The left hand side compares with whose integral is ln(cos 𝜃 + sin 𝜃) since it is of
cos 𝜃+sin 𝜃
𝑓 ′ (𝜃)
the form ∫ 𝑑𝜃 whose anti-derivative is ln[𝑓(𝜃)].
𝑓(𝜃)

Hence integration of equation (∗∗) yields ln[cos(𝑥 + 𝑦) + sin(𝑥 + 𝑦)] = 𝑥 − 𝑦 + ln 𝑐2 or

cos(𝑥 + 𝑦) + sin(𝑥 + 𝑦)
ln[cos(𝑥 + 𝑦) + sin(𝑥 + 𝑦)] − ln 𝑐2 = 𝑥 − 𝑦 𝑜𝑟 ln [ ] = 𝑥 − 𝑦 𝑖. 𝑒.
𝑐2

cos(𝑥 + 𝑦) + sin(𝑥 + 𝑦) cos(𝑥 + 𝑦) + sin(𝑥 + 𝑦)

= 𝑒 𝑥−𝑦 𝑜𝑟 = 𝑐2
𝑐2 𝑒 𝑥−𝑦

Therefore [cos(𝑥 + 𝑦) + sin(𝑥 + 𝑦)]𝑒 −(𝑥−𝑦) = 𝑐2

𝜋
tan(𝑥+𝑦+ )
Thus the general solution is given by∅ [ 4
, [cos(𝑥 + 𝑦) + sin(𝑥 + 𝑦)]𝑒 −(𝑥−𝑦) ] = 0.
𝑧 √2

26
Integral Surfaces Passing Through a Given Curve

The general solution for a linear partial differential equation can be obtained by solving the
𝑑𝑥 𝑑𝑦 𝑑𝑧
auxiliary system = = .
𝑃 𝑄 𝑅

The general solution can be used to find an integral surface containing a given curve.

Suppose the solutions to the auxiliary system are:

𝑢(𝑥, 𝑦, 𝑧) = 𝑐1
} … (1)
𝑣(𝑥, 𝑦, 𝑧) = 𝑐2

Then the solution of the given partial differential equation can be written in the form

∅(𝑢, 𝑣) = 0 … (2)

Suppose we wish to determine an integral surface containing a given curve C described by the
parametric equations

𝑥 = 𝑥(𝑡), 𝑦 = 𝑦(𝑡), 𝑧 = 𝑧(𝑡) … (3)

where t is a parameter. Then, the particular solution of (1) must be

𝑢[𝑥(𝑡), 𝑦(𝑡), 𝑧(𝑡)] = 𝑐1

} … (4)
𝑣[𝑥(𝑡), 𝑦(𝑡), 𝑧(𝑡)] = 𝑐2

Thus, we have two relations from which we can eliminate the parameter t to obtain a relation of
the type

∅(𝑐1 , 𝑐2 ) = 0 … (5)

Which leads to the solution given by equation (2) containing the curve C.

Example 1

Find the integral surface of the linear partial differential equation

𝑥(𝑦 2 + 𝑧)𝑝 − 𝑦(𝑥 2 + 𝑧)𝑞 = (𝑥 2 − 𝑦 2 )𝑧

containing the straight line 𝑥 + 𝑦 = 0, 𝑧 = 1.

The auxiliary equations for the given PDE are

𝑑𝑥 𝑑𝑦 𝑑𝑧
= = 2
𝑥(𝑦 + 𝑧) −𝑦(𝑥 + 𝑧) (𝑥 − 𝑦 2 )𝑧
2 2

Using the multipliers yz, xz and xy, we have

27
 𝑦𝑧𝑑𝑥 𝑥𝑧𝑑𝑦 𝑥𝑦𝑑𝑧
= =
𝑥𝑦𝑧(𝑦 + 𝑧) −𝑥𝑦𝑧(𝑥 + 𝑧) 𝑥𝑦𝑧(𝑥 2 − 𝑦 2 )
2 2

𝑦𝑧𝑑𝑥 + 𝑥𝑧𝑑𝑦 + 𝑥𝑦𝑑𝑧 𝑑𝑥

𝑜𝑟 =
𝑥𝑦𝑧[(𝑦 2 + 𝑧) − (𝑥 + 𝑧) + (𝑥 − 𝑦 )] 𝑥(𝑦 2 + 𝑧)
2 2 2

𝑖. 𝑒 𝑦𝑧𝑑𝑥 + 𝑥𝑧𝑑𝑦 + 𝑥𝑦𝑑𝑧 = 0 𝑜𝑟 𝑑(𝑥𝑦𝑧) = 0

which on integration gives 𝑥𝑦𝑧 = 𝑐1 … (1).

Using multipliers x, y and – 1 , we have

𝑥𝑑𝑥 + 𝑦𝑑𝑦 − 𝑑𝑧 𝑑𝑧
= 2
𝑥 2 (𝑦 2 + 𝑧) − 𝑦 (𝑥 + 𝑧) − 𝑧(𝑥 − 𝑦 ) (𝑥 − 𝑦 2 )𝑧
2 2 2 2

𝑥𝑑𝑥 + 𝑦𝑑𝑦 − 𝑑𝑧 𝑑𝑥
𝑜𝑟 =
(𝑥 2 − 𝑦 )𝑧 − 𝑧(𝑥 − 𝑦 ) 𝑥(𝑦 2 + 𝑧)
2 2 2

𝑖. 𝑒. 𝑥𝑑𝑥 + 𝑦𝑑𝑦 − 𝑑𝑧 = 0

𝑥2 𝑦2
𝑜𝑟 + − 𝑧 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑜𝑟 𝑥 2 + 𝑦 2 − 2𝑧 = 𝑐2 … (2)
2 2
Equation (1) and (2) lead to a general solution ∅(𝑥𝑦𝑧, 𝑥 2 + 𝑦 2 − 2𝑧) = 0. We seek the integral
surface containing the straight line 𝑥 + 𝑦 = 0, 𝑧 = 1.

For the line in question, we have the equation in parametric form as 𝑥 = 𝑡, 𝑦 = −𝑡 𝑎𝑛𝑑 𝑧 = 1.

Substituting these values in equations (1) and (2), we obtain

−𝑡 2 = 𝑐1
} … (3)
𝑡 2 − 1 = 𝑐2

Eliminating parameter t from (3) gives −1 = 𝑐1 + 𝑐2 𝑜𝑟 𝑐1 + 𝑐2 + 1 = 0.

Hence the required integral surface is 𝑥𝑦𝑧 + 𝑥 2 + 𝑦 2 − 2𝑧 + 1 = 0

𝑜𝑟 𝑥 2 + 𝑦 2 + 𝑥𝑦𝑧 − 2𝑧 + 1 = 0

Example 2

Find the integral surface of the linear PDE 𝑥𝑝 + 𝑦𝑞 = 𝑧 which contains the circle defined by

𝑥 2 + 𝑦 2 + 𝑧 2 = 4, 𝑥 + 𝑦 + 𝑧 = 2
𝑑𝑥 𝑑𝑦 𝑑𝑧
The auxiliary system of the given PDE is = = .
𝑥 𝑦 𝑧

28
 𝑥
Integration of the first two members gives ln 𝑥 = ln 𝑦 + 𝑐 𝑜𝑟 ln 𝑦 = 𝑐

𝑥
𝑖. 𝑒. = 𝑐1 … (1)
𝑦
𝑦
Similarly, integration of the last two members yields = 𝑐2 … (2)
𝑧

𝑥 𝑦
Hence the general solution of the given PDE is ∅ (𝑦 , 𝑧 ) = 0. If this general solution also
𝑥 𝑦
contains the given circle, then we have to find a relation between 𝑦 and 𝑧 .

𝑥2 + 𝑦2 + 𝑧2 = 4
The equation of the circle is } … (∗)
𝑥+𝑦+𝑧 = 2

From equations (1) and (2), we have

𝑥 𝑦 𝑥
𝑦= ,𝑧 = =
𝑐1 𝑐2 𝑐1 𝑐2

Substituting these values of y and z in (*), we find

𝑥2 𝑥2 1 1
𝑥 + 2 + 2 2 = 4 𝑜𝑟 𝑥 2 (1 + 2 + 2 2 ) = 4
2
𝑐1 𝑐1 𝑐2 𝑐1 𝑐1 𝑐2

𝑥 𝑥 1 1
𝑎𝑛𝑑 𝑥 + + = 2 𝑜𝑟 𝑥 (1 + + )=2
𝑐1 𝑐1 𝑐2 𝑐1 𝑐1 𝑐2

Hence we observe that

1 1 1 1 2 1 1 1 1
1 + 2 + 2 2 = [1 + + ] = [1 + + ] [1 + + ]
𝑐1 𝑐1 𝑐2 𝑐1 𝑐1 𝑐2 𝑐1 𝑐1 𝑐2 𝑐1 𝑐1 𝑐2

1 1 1 1 1 1 1 1 1 1
𝑜𝑟 1 + 2
+ 2 2 =1+ + + + 2+ 2 + + 2 + 2 2
𝑐1 𝑐1 𝑐2 𝑐1 𝑐1 𝑐2 𝑐1 𝑐1 𝑐1 𝑐2 𝑐1 𝑐2 𝑐1 𝑐2 𝑐1 𝑐2

1 1 2 2 1 2 1
𝑖. 𝑒. 1 + + = 1 + + + + +
𝑐1 2 𝑐1 2 𝑐2 2 𝑐1 𝑐1 𝑐2 𝑐1 2 𝑐1 2 𝑐2 𝑐1 2 𝑐2 2
2 2 2 2 1 1
which on simplification gives 𝑐 + 𝑐 +𝑐 2
= 0 𝑜𝑟 𝑐 [1 + 𝑐 + 𝑐 ]=0
1 1 𝑐2 1 𝑐2 1 2 1 𝑐2

1 1
𝑜𝑟 1 + + =0
𝑐2 𝑐1 𝑐2

𝑖. 𝑒. 𝑐1 𝑐2 + 𝑐1 + 1 = 0

29
 𝑥 𝑦
Replacing 𝑐1 𝑏𝑦 𝑎𝑛𝑑 𝑐2 𝑏𝑦 gives the required integral surface i.e.
𝑦 𝑧

𝑥 𝑦 𝑥 𝑥 𝑥
. + + 1 = 0 𝑜𝑟 + + 1 = 0 𝑜𝑟 𝑥𝑦 + 𝑥𝑧 + 𝑦𝑧 = 0.
𝑦 𝑧 𝑦 𝑧 𝑦

Example 3

Find the integral surface of the partial differential equation 𝑥 3 𝑝 + 𝑦(3𝑥 2 + 𝑦)𝑞 = 𝑧(2𝑥 2 + 𝑦)
passing through the curve 𝑥 = 1, 𝑧 = 𝑦 2 + 𝑦.

The PDE is Lagrange’s type 𝑃𝑝 + 𝑄𝑞 = 𝑅. So, the Lagrange’s auxiliary equation will be

𝑑𝑥 𝑑𝑦 𝑑𝑧
3
= =
𝑥 𝑦(3𝑥 + 𝑦) 𝑧(2𝑥 2 + 𝑦)
2

𝑑𝑥 𝑑𝑦
At first we take = 𝑦(3𝑥 2+𝑦).
𝑥3

3𝑥2
𝑑𝑦 𝑦(3𝑥 2 +𝑦) 1 𝑑𝑦 +1 31 1
𝑦
This is equivalent to 𝑑𝑥 = 𝑜𝑟 = = 𝑥 𝑦 + 𝑥 3 … (∗)
𝑥3 𝑦2 𝑑𝑥 𝑥 3

1
Put 𝑦 = 𝑦 −1 = 𝑡.

1 𝑑𝑦 𝑑𝑡 1 𝑑𝑦 𝑑𝑡 𝑑𝑡 3𝑡 1
Therefore − 𝑦 2 𝑑𝑥 = 𝑑𝑥 𝑜𝑟 = − 𝑑𝑥 and equation (*) becomes – 𝑑𝑥 = + 𝑥3
𝑦2 𝑑𝑥 𝑥

𝑑𝑡 3𝑡 1 3𝑑𝑥 3
𝑜𝑟 + = − 3 , 𝑤ℎ𝑜𝑠𝑒 𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑖𝑛𝑔 𝑓𝑎𝑐𝑡𝑜𝑟 𝑖𝑠 𝑒 ∫ 𝑥 = 𝑒 ln(𝑥 ) = 𝑥 3
𝑑𝑥 𝑥 𝑥
1 𝑥3
Thus 𝑡𝑥 3 = ∫ (− 𝑥 3 ) (𝑥 3 )𝑑𝑥 𝑜𝑟 𝑡𝑥 3 = ∫(−1) 𝑑𝑥 = −𝑥 + 𝑐1 𝑜𝑟 = −𝑥 + 𝑐1 .
𝑦

𝑥3
Hence + 𝑥 = 𝑐1 … (1)
𝑦

Next, from
𝑑𝑥 𝑑𝑦 𝑑𝑧
3
= = 𝑤𝑒 𝑡𝑎𝑘𝑒
𝑥 𝑦(3𝑥 + 𝑦) 𝑧(2𝑥 2 + 𝑦)
2

𝑧𝑑𝑦−𝑦𝑑𝑧 𝑧𝑑𝑦−𝑦𝑑𝑧 𝑑𝑥 𝑑𝑥 𝑧𝑑𝑦−𝑦𝑑𝑧 𝑑𝑦 𝑑𝑧

= = 𝑜𝑟 = = − .
𝑦𝑧(3𝑥 2 +𝑦)−𝑦𝑧(2𝑥 2 +𝑦) 𝑥 2 𝑦𝑧 𝑥3 𝑥 𝑦𝑧 𝑦 𝑧

𝑑𝑥 𝑑𝑦 𝑑𝑧 𝑑𝑥 𝑑𝑦 𝑑𝑧
Thus − + = 0 𝑜𝑟 ∫ −∫ +∫ = 0 𝑜𝑟 ln 𝑥 − ln 𝑦 + ln 𝑧 = ln 𝑐2
𝑥 𝑦 𝑧 𝑥 𝑦 𝑧

𝑜𝑟 ln(𝑥𝑧) − ln 𝑦 = ln 𝑐2
𝑥𝑧
𝑖. 𝑒 ln = ln 𝑐2
𝑦

30
 𝑥𝑧
Hence = 𝑐2 … (2).
𝑦

𝑥3 𝑥𝑧
Thus, the general solution of the given PDE is ∅ ( 𝑦 + 𝑥, 𝑦 ) = 0.

We are given the curve 𝑥 = 1, 𝑧 = 𝑦 2 + 𝑦. Putting the values in the equations (1) and (2) we get

1
𝑐1 = + 1 … (3)
𝑦

𝑐2 = (𝑦 + 1) … (4)

Putting the value of y of equation (4) in equation (3) we get

(𝑐2 − 1)(𝑐1 − 1) = 1

𝑥𝑧 𝑥3
Hence the equation ( 𝑦 − 1) ( 𝑦 + 𝑥 − 1) = 1 which is the required integral surface.

SIMULTANEOUS DIFFERENTIAL EQUATIONS OF THE FIRST ORDER

Compatible Systems of First Order Equations

Two first order PDEs are said to be compatible if they have a common solution.

Let 𝑓(𝑥, 𝑦, 𝑧, 𝑝, 𝑞) = 0 𝑎𝑛𝑑 𝑔(𝑥, 𝑦, 𝑧, 𝑝, 𝑞) = 0 be two first order partial differential equations.

The necessary and sufficient conditions for the two PDEs to be compatible is

𝜕(𝑓, 𝑔) 𝜕(𝑓, 𝑔) 𝜕(𝑓, 𝑔) 𝜕(𝑓, 𝑔)

+𝑝 + +𝑞 =0
𝜕(𝑥, 𝑝) 𝜕(𝑧, 𝑝) 𝜕(𝑦, 𝑞) 𝜕(𝑧, 𝑞)

Jacobian Matrix & Determinant

The Jacobian matrix is the matrix of all first order partial derivatives of a multi – valued
function. When the matrix is a square matrix, both the matrix and its determinant are referred to
as the Jacobian.
𝜕𝑓1 𝜕𝑓1
𝜕𝑥 𝜕𝑥2 𝜕(𝑓 ,𝑓 )
𝐽= | 𝜕𝑓1 𝜕𝑓2
| = 𝜕(𝑥1,𝑥2 )
2 1 2
𝜕𝑥1 𝜕𝑥2

Examples
𝑥𝑝 − 𝑦𝑞 = 𝑥
1.Show that the following PDEs } are compatible and hence find their common
𝑥 2 𝑝 + 𝑞 = 𝑥𝑧
solution.

31
Let 𝑓 = 𝑥𝑝 − 𝑦𝑞 − 𝑥 = 0 … (1)𝑎𝑛𝑑 𝑔 = 𝑥 2 𝑝 + 𝑞 − 𝑥𝑧 = 0 … (2). Then

𝜕(𝑓, 𝑔) (𝑝 − 1) 𝑥
=| | = 𝑝𝑥 2 − 𝑥 2 − 2𝑥 2 𝑝 + 𝑥𝑧 = 𝑥𝑧 − 𝑥 2 𝑝 − 𝑥 2
𝜕(𝑥, 𝑝) (2𝑥𝑝 − 𝑧) 𝑥2

𝜕(𝑓, 𝑔) 0 𝑥
=| | = 𝑥2
𝜕(𝑧, 𝑝) −𝑥 𝑥2

𝜕(𝑓, 𝑔) −𝑞 −𝑦
=| | = −𝑞
𝜕(𝑦, 𝑞) 0 1

𝜕(𝑓, 𝑔) 0 −𝑦
=| | = −𝑥𝑦
𝜕(𝑧, 𝑞) −𝑥 1

Thus,
𝜕(𝑓, 𝑔) 𝜕(𝑓, 𝑔) 𝜕(𝑓, 𝑔) 𝜕(𝑓, 𝑔)
+𝑝 + +𝑞
𝜕(𝑥, 𝑝) 𝜕(𝑧, 𝑝) 𝜕(𝑦, 𝑞) 𝜕(𝑧, 𝑞)

= 𝑥𝑧 − 𝑥 2 𝑝 − 𝑥 2 + 𝑝𝑥 2 − 𝑞 − 𝑞𝑥𝑦 = 𝑥𝑧 − 𝑞 − 𝑞𝑥𝑦 − 𝑥 2

= 𝑥𝑧 − 𝑞 − 𝑥(𝑞𝑦 + 𝑥) = 𝑥𝑧 − 𝑞 − 𝑥(𝑥𝑝), 𝑠𝑖𝑛𝑐𝑒 𝑞𝑦 + 𝑥 = 𝑥𝑝 𝑓𝑟𝑜𝑚 (1)

= 𝑥𝑧 − 𝑞 − 𝑥 2 𝑝

= 0, 𝑢𝑠𝑖𝑛𝑔 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 (2)

Hence the given PDEs are compatible.

Suppose that 𝑧 = 𝑓(𝑥, 𝑦) is a differentiable function of x and y which are expressed in

parametric form as 𝑥 = 𝑔(𝑡) 𝑎𝑛𝑑 𝑦 = ℎ(𝑡). Then, 𝑧 = 𝑓[𝑔(𝑡), ℎ(𝑡)].
𝑑𝑧 𝜕𝑧 𝑑𝑥 𝜕𝑧 𝑑𝑦
From chain rule of differentiation for a function of two variables, 𝑑𝑡 = 𝜕𝑥 𝑑𝑡 + 𝜕𝑦 𝑑𝑡 .

𝜕𝑧 𝜕𝑧
𝑖. 𝑒 𝑑𝑧 = 𝑑𝑥 + 𝑑𝑦 𝑜𝑟 𝑑𝑧 = 𝑝𝑑𝑥 + 𝑞𝑑𝑦 … (3)
𝜕𝑥 𝜕𝑦

The solution to the compatible system can be obtain from equation (3) by substituting for p and q

Now, solving equations (1) and (2) for p and q we obtain

(1) → 𝑥𝑝 − 𝑦𝑞 − 𝑥 = 0

𝑦(2) → 𝑥 2 𝑦𝑝 + 𝑦𝑞 − 𝑥𝑦𝑧 = 0

𝐴𝑑𝑑𝑖𝑛𝑔, 𝑥𝑝 + 𝑥 2 𝑦𝑝 − 𝑥 − 𝑥𝑦𝑧 = 0

𝑖. 𝑒. 𝑝(𝑥 + 𝑥 2 𝑦) = 𝑥 + 𝑥𝑦𝑧 𝑜𝑟 𝑝𝑥(1 + 𝑥𝑦) = 𝑥(1 + 𝑦𝑧)

32
 𝑥(1 + 𝑦𝑧) 1 + 𝑦𝑧
𝑖. 𝑒 𝑝 = 𝑜𝑟 𝑝 = .
𝑥(1 + 𝑥𝑦) 1 + 𝑥𝑦

Again,

𝑥(1) → 𝑥 2 𝑝 − 𝑥𝑦𝑞 − 𝑥 2 = 0

(2) → 𝑥 2 𝑝 + 𝑞 − 𝑥𝑧 = 0

𝑆𝑢𝑏𝑡𝑟𝑎𝑐𝑡𝑖𝑛𝑔, −𝑥𝑦𝑞 − 𝑞 − 𝑥 2 + 𝑥𝑧 = 0

𝑖. 𝑒. 𝑞 + 𝑥𝑦𝑞 = 𝑥𝑧 − 𝑥 2 𝑜𝑟 𝑞(1 + 𝑥𝑦) = 𝑥(𝑧 − 𝑥)

𝑥(𝑧 − 𝑥)
𝑖. 𝑒 𝑞 =
1 + 𝑥𝑦

In order to get the solution of the given system, we integrate the equation

𝑑𝑧 = 𝑝𝑑𝑥 + 𝑞𝑑𝑦

1 + 𝑦𝑧 𝑥(𝑧 − 𝑥)
𝑖. 𝑒 𝑑𝑧 = 𝑑𝑥 + 𝑑𝑦
1 + 𝑥𝑦 1 + 𝑥𝑦

Rearranging terms in the equation yields

1 + 𝑦𝑧 𝑥(𝑧 − 𝑥)
𝑑𝑧 − 𝑑𝑥 = 𝑑𝑥 − 𝑑𝑥 + 𝑑𝑦
1 + 𝑥𝑦 1 + 𝑥𝑦

(1 + 𝑦𝑧)𝑑𝑥 − (1 + 𝑥𝑦)𝑑𝑥 𝑥(𝑧 − 𝑥)

𝑜𝑟 𝑑𝑧 − 𝑑𝑥 = + 𝑑𝑦
1 + 𝑥𝑦 1 + 𝑥𝑦

𝑦(𝑧 − 𝑥) 𝑥(𝑧 − 𝑥) 𝑑𝑧 − 𝑑𝑥 𝑦𝑑𝑥 + 𝑥𝑑𝑦

𝑖. 𝑒. 𝑑𝑧 − 𝑑𝑥 = 𝑑𝑥 + 𝑑𝑦 𝑜𝑟 =
1 + 𝑥𝑦 1 + 𝑥𝑦 𝑧−𝑥 1 + 𝑥𝑦
𝑑(𝑧−𝑥) 𝑑(1+𝑥𝑦)
Thus, = , and integration gives ln(𝑧 − 𝑥) = ln(1 + 𝑥𝑦) + ln 𝑐 = ln[𝑐(1 + 𝑥𝑦)] .
𝑧−𝑥 1+𝑥𝑦

𝑖. 𝑒. 𝑧 − 𝑥 = 𝑐(1 + 𝑥𝑦)

Hence the solution of the system is found to be 𝑧 = 𝑥 + 𝑐(1 + 𝑥𝑦).

2.Show that the PDEs 𝑥𝑝 = 𝑦𝑞 𝑎𝑛𝑑 𝑧(𝑥𝑝 + 𝑦𝑞) = 2𝑥𝑦 are compatible and hence find the
solution.

Let 𝑓 = 𝑥𝑝 − 𝑦𝑞 = 0 … (1)𝑎𝑛𝑑 𝑔 = 𝑥𝑧𝑝 + 𝑦𝑧𝑞 − 2𝑥𝑦 = 0 … (2)

𝜕(𝑓, 𝑔) 𝑝 𝑥
= |(𝑧𝑝 − 2𝑦) 𝑥𝑧| = 𝑝𝑥𝑧 − 𝑝𝑥𝑧 + 2𝑥𝑦 = 2𝑥𝑦
𝜕(𝑥, 𝑝)
33
 𝜕(𝑓, 𝑔) 0 𝑥
=| | = −𝑥 2 𝑝 − 𝑥𝑦𝑞
𝜕(𝑧, 𝑝) 𝑥𝑝 + 𝑦𝑞 𝑥𝑧

𝜕(𝑓, 𝑔) −𝑞 −𝑦
= |𝑧𝑞 − 2𝑥 𝑦𝑧 | = −𝑦𝑧𝑞 + 𝑦𝑧𝑞 − 2𝑥𝑦 = −2𝑥𝑦
𝜕(𝑦, 𝑞)

𝜕(𝑓, 𝑔) 0 −𝑦
=| | = 𝑥𝑦𝑝 + 𝑦 2 𝑞
𝜕(𝑧, 𝑞) 𝑥𝑝 + 𝑦𝑞 𝑦𝑧

Thus,
𝜕(𝑓, 𝑔) 𝜕(𝑓, 𝑔) 𝜕(𝑓, 𝑔) 𝜕(𝑓, 𝑔)
+𝑝 + +𝑞
𝜕(𝑥, 𝑝) 𝜕(𝑧, 𝑝) 𝜕(𝑦, 𝑞) 𝜕(𝑧, 𝑞)

= 2𝑥𝑦 − 𝑥 2 𝑝2 − 𝑥𝑦𝑝𝑞 − 2𝑥𝑦 + 𝑥𝑦𝑝𝑞 + 𝑦 2 𝑞 2 = 𝑦 2 𝑞 2 − 𝑥 2 𝑝2 = 0, 𝑢𝑠𝑖𝑛𝑔 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 (1).

Thus, the given PDEs are compatible.

Solving equations (1) and (2) for p and q, we have

(2) → 𝑥𝑧𝑝 + 𝑦𝑧𝑞 − 2𝑥𝑦 = 0

𝑧(1) → 𝑥𝑧𝑝 − 𝑦𝑧𝑞 = 0

𝐴𝑑𝑑𝑖𝑛𝑔, 2𝑥𝑧𝑝 − 2𝑥𝑦 = 0

2𝑥𝑦 𝑦
𝑖. 𝑒 2𝑥𝑧𝑝 = 2𝑥𝑦 𝑜𝑟 𝑝 = =
2𝑥𝑧 𝑧
𝑥𝑝 𝑥𝑦 𝑥
From (1), 𝑦𝑞 = 𝑥𝑝 𝑜𝑟 𝑞 = = 𝑧𝑦 = 𝑧 . Therefore
𝑦

𝑥
𝑞= . 𝐵𝑢𝑡
𝑧
𝑦 𝑥
𝑑𝑧 = 𝑝𝑑𝑥 + 𝑞𝑑𝑦 𝑜𝑟 𝑑𝑧 = 𝑑𝑥 + 𝑑𝑦
𝑧 𝑧
𝑖. 𝑒. 𝑧𝑑𝑧 = 𝑦𝑑𝑥 + 𝑥𝑑𝑦 𝑜𝑟 𝑧𝑑𝑧 = 𝑑(𝑥𝑦). 𝑇ℎ𝑢𝑠 𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑖𝑜𝑛 𝑦𝑖𝑒𝑙𝑑𝑠

𝑧2
= 𝑥𝑦 + 𝑘 𝑜𝑟 𝑧 2 = 2𝑥𝑦 + 2𝑘. 𝐻𝑒𝑛𝑐𝑒
2
𝑧 2 = 2𝑥𝑦 + 𝑐 is the required solution.

Exercise

Show that the equations

𝑝2 + 𝑞 2 = 1 𝑎𝑛𝑑

34
 (𝑝2 + 𝑞 2 )𝑥 = 𝑝𝑧

are compatible and hence find the solution.

Let 𝑓 = 𝑝2 + 𝑞 2 − 1 = 0 … (1)𝑎𝑛𝑑 𝑔 = 𝑥𝑝2 + 𝑥𝑞 2 − 𝑝𝑧 = 0 … (2)

𝜕(𝑓, 𝑔) 0 2𝑝
=| 2 | = −2𝑝3 − 2𝑝𝑞 2
𝜕(𝑥, 𝑝) (𝑝 + 𝑞 2 ) (2𝑥𝑝 − 𝑧)

𝜕(𝑓, 𝑔) 0 2𝑝
=| | = 2𝑝2
𝜕(𝑧, 𝑝) −𝑝 (2𝑥𝑝 − 𝑧)

𝜕(𝑓, 𝑔) 0 2𝑞
=| |=0
𝜕(𝑦, 𝑞) 0 2𝑥𝑞

𝜕(𝑓, 𝑔) 0 2𝑞
=| | = 2𝑝𝑞
𝜕(𝑧, 𝑞) −𝑝 2𝑥𝑞

Thus,
𝜕(𝑓, 𝑔) 𝜕(𝑓, 𝑔) 𝜕(𝑓, 𝑔) 𝜕(𝑓, 𝑔)
+𝑝 + +𝑞
𝜕(𝑥, 𝑝) 𝜕(𝑧, 𝑝) 𝜕(𝑦, 𝑞) 𝜕(𝑧, 𝑞)

= −2𝑝3 − 2𝑝𝑞 2 + 2𝑝3 + 0 + 2𝑝𝑞 2 = 2𝑝𝑞 2 − 2𝑝𝑞 2 = 0

Hence the given PDEs are compatible.

Equations (1) and (2) are solved for p and q.

(2) → 𝑥𝑝2 + 𝑥𝑞 2 − 𝑝𝑧 = 0

𝑥(1) → 𝑥𝑝2 + 𝑥𝑞 2 − 𝑥 = 0

𝑆𝑢𝑏𝑡𝑟𝑎𝑐𝑡𝑖𝑛𝑔, −𝑝𝑧 + 𝑥 = 0
𝑥
𝑖. 𝑒. 𝑥 = 𝑝𝑧 𝑜𝑟 𝑝 =
𝑧
𝑥2 𝑧 2 −𝑥 2
From equation (1), 𝑞 2 = 1 − 𝑝2 . Substituting the value of p gives 𝑞 2 = 1 − 𝑧 2 = .
𝑧2

√𝑧 2 −𝑥 2
Therefore,𝑞 = .
𝑧

𝑥𝑑𝑥 √𝑧 2 −𝑥 2
Now, 𝑑𝑧 = 𝑝𝑑𝑥 + 𝑞𝑑𝑦 = + 𝑑𝑦 𝑜𝑟 𝑧𝑑𝑧 = 𝑥𝑑𝑥 + √𝑧 2 − 𝑥 2 𝑑𝑦.
𝑧 𝑧

𝑧𝑑𝑧 − 𝑥𝑑𝑥
𝑜𝑟 𝑧𝑑𝑧 − 𝑥𝑑𝑥 = √𝑧 2 − 𝑥 2 𝑑𝑦 𝑜𝑟 = 𝑑𝑦
√𝑧 2 − 𝑥 2

35
 1 1
Thus 𝑑 [(𝑧 2 − 𝑥 2 )2 ] = 𝑑𝑦, which on integration yields (𝑧 2 − 𝑥 2 )2 = 𝑦 + 𝑐

𝑜𝑟 𝑧 2 − 𝑥 2 = (𝑦 + 𝑐)2

Therefore 𝑧 2 = 𝑥 2 + (𝑦 + 𝑐)2 is the required solution.

Non – linear Partial Differential Equations

Non – linear Partial Differential Equations can be solved using various methods

CHARPIT’S METHOD

Charpit’s Method is one of the general methods for finding the complete integral or complete
solution of a non –linear PDE of first order of the form 𝑓(𝑥, 𝑦, 𝑧, 𝑝, 𝑞) = 0 … (1)

The basic principle behind the method is the consideration of existence / introduction of another
PDE of first order of the form 𝑔(𝑥, 𝑦, 𝑧, 𝑝, 𝑞) = 0 … (2), which is compatible with (1).

Equations (1) and (2) are then solved for p and q and substitution made in the equation

𝑑𝑧 = 𝑝(𝑥, 𝑦, 𝑧)𝑑𝑥 + 𝑞(𝑥, 𝑦, 𝑧)𝑑𝑦 … (3)

The solution of equation (3), if it exists is the complete integral of (1).

The necessary and sufficient condition for compatibility of equations (1) and (2) is

𝜕(𝑓, 𝑔) 𝜕(𝑓, 𝑔) 𝜕(𝑓, 𝑔) 𝜕(𝑓, 𝑔)

+𝑝 + +𝑞 = 0 … (4)
𝜕(𝑥, 𝑝) 𝜕(𝑧, 𝑝) 𝜕(𝑦, 𝑞) 𝜕(𝑧, 𝑞)

On expansion we have

𝜕𝑓 𝜕𝑔 𝜕𝑓 𝜕𝑔 𝜕𝑓 𝜕𝑔 𝜕𝑓 𝜕𝑔 𝜕𝑓 𝜕𝑔 𝜕𝑓 𝜕𝑔 𝜕𝑓 𝜕𝑔 𝜕𝑓 𝜕𝑔
( − )+𝑝( − )+( − )+𝑞( − )=0
𝜕𝑥 𝜕𝑝 𝜕𝑝 𝜕𝑥 𝜕𝑧 𝜕𝑝 𝜕𝑝 𝜕𝑧 𝜕𝑦 𝜕𝑞 𝜕𝑞 𝜕𝑦 𝜕𝑧 𝜕𝑞 𝜕𝑞 𝜕𝑧

which can be written as

𝜕𝑓 𝜕𝑔 𝜕𝑓 𝜕𝑔 𝜕𝑓 𝜕𝑔 𝜕𝑓 𝜕𝑔 𝜕𝑓 𝜕𝑔 𝜕𝑓 𝜕𝑔 𝜕𝑓 𝜕𝑔 𝜕𝑓 𝜕𝑔
( − )+𝑝( − )+( − )+𝑞( − ) = 0 𝑜𝑟
𝜕𝑝 𝜕𝑥 𝜕𝑥 𝜕𝑝 𝜕𝑝 𝜕𝑧 𝜕𝑧 𝜕𝑝 𝜕𝑞 𝜕𝑦 𝜕𝑦 𝜕𝑞 𝜕𝑞 𝜕𝑧 𝜕𝑧 𝜕𝑞

𝜕𝑔 𝜕𝑔 𝜕𝑔 𝜕𝑔 𝜕𝑔
𝑓𝑝 + 𝑓𝑞 + (𝑝𝑓𝑝 + 𝑞𝑓𝑞 ) − (𝑓𝑥 + 𝑝𝑓𝑧 ) − (𝑓𝑦 + 𝑞𝑓𝑧 ) = 0 … (5)
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑝 𝜕𝑞

Equation (5) is a linear PDE, from which we can determine 𝑔.

𝜕𝑧 𝜕𝑧
{𝐼𝑡 𝑐𝑜𝑚𝑝𝑎𝑟𝑒𝑠 𝑤𝑖𝑡ℎ 𝑃𝑝 + 𝑄𝑞 = 𝑅 𝑜𝑟𝑃 + 𝑄 =
𝜕𝑥 𝜕𝑦
𝑑𝑥 𝑑𝑦 𝑑𝑧
𝑅 𝑓𝑟𝑜𝑚 𝑤ℎ𝑖𝑐ℎ 𝑤𝑒 𝑓𝑜𝑟𝑚 𝑡ℎ𝑒 𝑐𝑜𝑟𝑟𝑒𝑠𝑝𝑜𝑛𝑑𝑖𝑛𝑔 𝑎𝑢𝑥𝑖𝑙𝑖𝑎𝑟𝑦 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛𝑠 = = }.
𝑃 𝑄 𝑅

36
The auxiliary equations of (5) are

𝑑𝑥 𝑑𝑦 𝑑𝑧 𝑑𝑝 𝑑𝑞
= = = = … (6)
𝑓𝑝 𝑓𝑞 𝑝𝑓𝑝 + 𝑞𝑓𝑞 −(𝑓𝑥 + 𝑝𝑓𝑧 ) −(𝑓𝑦 + 𝑞𝑓𝑧 )

These equations are called Charpit’s equations. Any integral of equation (6) involving p or q or
both can be taken as the second relation of equation (2).

Then, the integration of equation (3) gives the complete integral as desired. It may be noted that
all Charpit’s equations need not be used, but it is enough to choose the simplest of them.

Examples

1.Find the complete integral of 𝑥 2 𝑝2 + 𝑦 2 𝑞 2 − 4 = 0 using Charpit’s method.

Let 𝑓 = 𝑥 2 𝑝2 + 𝑦 2 𝑞 2 − 4 = 0. The Charpit’s equations for the given PDE can be written as

𝑑𝑥 𝑑𝑦 𝑑𝑧 𝑑𝑝 𝑑𝑞
2
= 2 = 2 2 2 2
= 2
=
2𝑥 𝑝 2𝑦 𝑞 2(𝑥 𝑝 + 𝑦 𝑞 ) −2𝑥𝑝 −2𝑦𝑞 2

Considering the first and last but one terms, we have

𝑑𝑥 𝑑𝑝 𝑑𝑥 𝑑𝑝
2
= 2
𝑜𝑟 + =0
2𝑥 𝑝 −2𝑥𝑝 𝑥 𝑝

On integration, we get ln x + ln p = ln a or ln(𝑥𝑝) = ln 𝑎 𝑖. 𝑒 𝑥𝑝 = 𝑎.

From the given PDE, 𝑦 2 𝑞 2 = 4 − 𝑥 2 𝑝2 . Substituting the above result gives

2 2 2
4 − 𝑎2
2
√4 − 𝑎2
𝑦 𝑞 = 4 − 𝑎 𝑜𝑟 𝑞 = 𝑖. 𝑒 𝑞 =
𝑦2 𝑦
𝑑𝑥 𝑑𝑦
Substituting p and q in 𝑑𝑧 = 𝑝𝑑𝑥 + 𝑞𝑑𝑦 𝑤𝑒 𝑓𝑖𝑛𝑑 𝑡ℎ𝑎𝑡 𝑑𝑧 = 𝑎 + √4 − 𝑎2 .
𝑥 𝑦

On integration, the complete integral is found to be 𝑧 = 𝑎 ln 𝑥 + √4 − 𝑎2 ln 𝑦 + 𝑏.

2.Use Charpit’s method to solve the non – linear PDE 𝑞 = −𝑥𝑝 + 𝑝2 .

Let 𝑓 = 𝑝2 − 𝑥𝑝 − 𝑞 = 0. Thus

𝑓𝑥 = −𝑝, 𝑓𝑦 = 0, 𝑓𝑧 = 0, 𝑓𝑝 = 2𝑝 − 𝑥, 𝑓𝑞 = −1, 𝑓𝑥 + 𝑝𝑓𝑧 = −𝑝, 𝑓𝑦 + 𝑞𝑓𝑧 = 0

Hence the auxiliary equations are

𝑑𝑥 𝑑𝑦 𝑑𝑧 𝑑𝑝 𝑑𝑞
= = = =
2𝑝 − 𝑥 −1 𝑝(2𝑝 − 𝑥) + 𝑞(−1) 𝑝 −(𝑓𝑦 + 𝑞𝑓𝑧 )

37
 𝑑𝑝 𝑑𝑦 𝑝
From = −1 , 𝑤𝑒 ℎ𝑎𝑣𝑒 ln 𝑝 = −𝑦 + ln 𝑎 𝑜𝑟 ln 𝑝 − ln 𝑎 = −𝑦 𝑖. 𝑒. ln 𝑎 = −𝑦 𝑜𝑟 𝑝 = 𝑎𝑒 −𝑦 .
𝑝

From the PDE given, 𝑞 = −𝑥𝑝 + 𝑝2 𝑜𝑟 𝑞 = −𝑥𝑎𝑒 −𝑦 + 𝑎2 𝑒 −2𝑦 . Hence

𝑑𝑧 = 𝑝𝑑𝑥 + 𝑞𝑑𝑦 𝑏𝑒𝑐𝑜𝑚𝑒𝑠 𝑑𝑧 = 𝑎𝑒 −𝑦 𝑑𝑥 + (−𝑥𝑎𝑒 −𝑦 + 𝑎2 𝑒 −2𝑦 )𝑑𝑦

𝑜𝑟 𝑑𝑧 = 𝑎𝑒 −𝑦 𝑑𝑥 − 𝑎𝑥𝑒 −𝑦 𝑑𝑦 + 𝑎2 𝑒 −2𝑦 𝑑𝑦 𝑖. 𝑒. 𝑑𝑧 = 𝑑(𝑎𝑥𝑒 −𝑦 ) + 𝑎2 𝑒 −2𝑦 𝑑𝑦

1
Upon integration, 𝑧 = 𝑎𝑥𝑒 −𝑦 − 2 𝑎2 𝑒 −2𝑦 + 𝑏, which is the required solution.

Exercise

Use charpit’s method to solve the following non – linear PDEs

𝑧2
1.(𝑝2 + 𝑞 2 )𝑦 = 𝑞𝑧 𝐴𝑛𝑠 (𝑥 + 𝑏)2 + 𝑦 2 = 𝑎

Let 𝑓 = (𝑝2 + 𝑞 2 )𝑦 − 𝑞𝑧 = 0. Then

𝑓𝑥 = 0, 𝑓𝑦 = 𝑝2 + 𝑞 2 , 𝑓𝑧 = −𝑞, 𝑓𝑝 = 2𝑝𝑦, 𝑓𝑞 = 2𝑞𝑦 − 𝑧

The auxiliary equations are

𝑑𝑥 𝑑𝑦 𝑑𝑧 𝑑𝑝 𝑑𝑞
= = = =
𝑓𝑝 𝑓𝑞 𝑝𝑓𝑝 + 𝑞𝑓𝑞 −(𝑓𝑥 + 𝑝𝑓𝑧 ) −(𝑓𝑦 + 𝑞𝑓𝑧 )

𝑑𝑥 𝑑𝑦 𝑑𝑧 𝑑𝑝 𝑑𝑞
𝑖. 𝑒. = = 2 = =
2𝑝𝑦 2𝑞𝑦 − 𝑧 2𝑝 𝑦 + 2𝑞 𝑦 − 𝑞𝑧 𝑝𝑞 −[(𝑝 + 𝑞 2 ) − 𝑞 2 ]
2 2

𝑑𝑝 𝑑𝑞 𝑑𝑝 𝑑𝑞
From the last two members, 𝑝𝑞 = −𝑝2 𝑜𝑟 = −𝑝 𝑜𝑟 𝑝𝑑𝑝 + 𝑞𝑑𝑞 = 0.
𝑞

Integrating, 𝑝2 + 𝑞 2 = 𝑎(𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡).
𝑦
Substituting in (𝑝2 + 𝑞 2 )𝑦 − 𝑞𝑧 = 0, we have 𝑎𝑦 − 𝑞𝑧 = 0 𝑜𝑟 𝑞 = 𝑎( 𝑧 ).

𝑦 𝑎𝑦
Thus 𝑝2 + 𝑎2 ( 𝑧 )2 = 𝑎 𝑜𝑟 𝑝2 = 𝑎 − ( 𝑧 )2. Therefore,

𝑎𝑦 2 𝑎𝑧 2 − 𝑎2 𝑦 2
𝑝 = √𝑎 − ( ) = √
𝑧 𝑧2

Substituting in 𝑑𝑧 = 𝑝𝑑𝑥 + 𝑞𝑑𝑦, we get

𝑎𝑧 2 − 𝑎2 𝑦 2 𝑎𝑦
𝑑𝑧 = √ 𝑑𝑥 + 𝑑𝑦 𝑜𝑟 𝑧𝑑𝑧 − 𝑎𝑦𝑑𝑦 = √𝑎𝑧 2 − 𝑎2 𝑦 2 𝑑𝑥
𝑧2 𝑧

38
 1
𝑧𝑑𝑧 − 𝑎𝑦𝑑𝑦 𝑎𝑧𝑑𝑧 − 𝑎2 𝑦𝑑𝑦 𝑑(𝑎𝑧 2 − 𝑎2 𝑦 2 )2
𝑜𝑟 1 = 𝑑𝑥 𝑜𝑟 1 = 𝑑𝑥 𝑜𝑟 = 𝑑𝑥
𝑎
(𝑎𝑧 2 − 𝑎 2 𝑦 2 )2 𝑎(𝑎𝑧 2 − 𝑎2 𝑦 2 )2

√𝑎𝑧 2 −𝑎2 𝑦 2 𝑎𝑧 2 −𝑎2 𝑦 2 𝑧2

Integrating, = 𝑥 + 𝑏 𝑜𝑟 (𝑥 + 𝑏)2 = 𝑜𝑟 (𝑥 + 𝑏)2 = − 𝑦2
𝑎 𝑎2 𝑎

𝑧2
Hence the complete integral is (𝑥 + 𝑏)2 + 𝑦 2 = .
𝑎

(𝑥−𝑎)2 (𝑦−𝑏)2
2.16𝑝2 𝑧 2 + 9𝑞 2 𝑧 2 + 4𝑧 2 − 4 = 0 𝐴𝑛𝑠 + 9 + 𝑧2 = 1
4
4

Let 𝑓 = 16𝑝2 𝑧 2 + 9𝑞 2 𝑧 2 + 4𝑧 2 − 4 = 0. Then

𝑓𝑥 = 0 = 𝑓𝑦 , 𝑓𝑧 = 32𝑝2 𝑧 + 18𝑞 2 𝑧 + 8𝑧, 𝑓𝑝 = 32𝑝𝑧 2 , 𝑓𝑞 = 18𝑞𝑧 2

Hence the auxiliary system

𝑑𝑥 𝑑𝑦 𝑑𝑧 𝑑𝑝
= = =
32𝑝𝑧 2 18𝑞𝑧 2 32𝑝2 𝑧 2 + 18𝑞 2 𝑧 2 −(0 + 32𝑝3 𝑧 + 18𝑝𝑞 2 𝑧 + 8𝑝𝑧)
𝑑𝑞
=
−(0 + 32𝑝2 𝑞𝑧 + 18𝑞 3 𝑧 + 8𝑞𝑧)

𝑑𝑝 𝑑𝑞 𝑑𝑥 𝑑𝑦
𝑜𝑟 = = =
32𝑝3 𝑧 2 2 3
+ 18𝑝𝑞 𝑧 + 8𝑝𝑧 32𝑝 𝑞𝑧 + 18𝑞 𝑧 + 8𝑞𝑧 −32𝑝𝑧 2 −18𝑞𝑧 2
𝑑𝑧
=
−32𝑝 𝑧 − 18𝑞 2 𝑧 2
2 2

Using the multipliers 4𝑧, 0,1,0,4𝑝, we find that

4𝑧𝑑𝑝 𝑑𝑥 4𝑝𝑑𝑧
= =
4𝑧(32𝑝3 𝑧 2
+ 18𝑝𝑞 𝑧 + 8𝑝𝑧) −32𝑝𝑧 2 4𝑝(−32𝑝2 𝑧 2 − 18𝑞 2 𝑧 2 )

4𝑧𝑑𝑝 + 𝑑𝑥 + 4𝑝𝑑𝑧 𝑑𝑦
𝑜𝑟 = , 𝑠𝑎𝑦
4𝑧(32𝑝3 𝑧 + 18𝑝𝑞 2 𝑧 + 8𝑝𝑧) − 32𝑝𝑧 2 + 4𝑝(−32𝑝2 𝑧 2 − 18𝑞 2 𝑧 2 ) −18𝑞𝑧 2

Thus, 𝑑𝑥 + 4𝑝𝑑𝑧 + 4𝑧𝑑𝑝 = 0 𝑜𝑟 𝑑𝑥 + 𝑑(4𝑝𝑧) = 0.

𝑎−𝑥 (𝑥−𝑎)
Therefore, 𝑥 + 4𝑝𝑧 = 𝑎 𝑜𝑟 𝑝 = 𝑜𝑟 𝑝 = − .
4𝑧 4𝑧

Substituting for p in the given differential equation gives

(𝑥 − 𝑎)2 2
16 𝑧 + 9𝑞 2 𝑧 2 + 4𝑧 2 − 4 = 0 𝑜𝑟 (𝑥 − 𝑎)2 + 9𝑞 2 𝑧 2 + 4𝑧 2 − 4 = 0
16𝑧 2
9𝑞 2 𝑧 2 = 4 − 4𝑧 2 − (𝑥 − 𝑎)2

39
 2
4 2
(𝑥 − 𝑎)2
𝑞 = 2 (1 − 𝑧 − )
9𝑧 4

2 1
𝑖. 𝑒. 𝑞 = √1 − 𝑧 2 − (𝑥 − 𝑎)2
3𝑧 4

Substituting into 𝑑𝑧 = 𝑝𝑑𝑥 + 𝑞𝑑𝑦, gives

(𝑥 − 𝑎) 2 1
𝑑𝑧 = − 𝑑𝑥 + √1 − 𝑧 2 − (𝑥 − 𝑎)2 𝑑𝑦
4𝑧 3𝑧 4

(𝑥 − 𝑎) 2 1 4𝑧𝑑𝑧 + (𝑥 − 𝑎)𝑑𝑥
𝑜𝑟 𝑑𝑧 + 𝑑𝑥 = √1 − 𝑧 2 − (𝑥 − 𝑎)2 𝑑𝑦 =
4𝑧 3𝑧 4 4𝑧

1
4𝑧𝑑𝑧+(𝑥−𝑎)𝑑𝑥 3𝑧 3[𝑧𝑑𝑧+ (𝑥−𝑎)𝑑𝑥]
4
Therefore, 𝑑𝑦 = . 𝑜𝑟 𝑑𝑦 =
4𝑧 1 1
2√1−𝑧 2 − (𝑥−𝑎)2 2√1−𝑧 2 − (𝑥−𝑎)2
4 4

1
3 1 2
𝑜𝑟 𝑑𝑦 = − 𝑑 {[1 − 𝑧 2 − (𝑥 − 𝑎)2 ] }
2 4

3 1
Integrating, 𝑦 − 𝑏 = − 2 √1 − 𝑧 2 − 4 (𝑥 − 𝑎)2

9
2 2
1 2
(𝑦 − 𝑏)2 1
𝑜𝑟 (𝑦 − 𝑏) = (1 − 𝑧 − (𝑥 − 𝑎) ) 𝑜𝑟 = 1 − 𝑧 2 − (𝑥 − 𝑎)2
4 4 9 4
4
(𝑥 − 𝑎)2 (𝑦 − 𝑏)2
𝑖. 𝑒. + + 𝑧 2 = 1 𝑖𝑠 𝑡ℎ𝑒 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛.
4 9
4

JACOBI’S METHOD

Jacobi’s method, like charpit’s method is used to solve non – linear partial differential equations.

The advantage of Jacobi’s method over Charpit’s method is that it can easily be extended to any
number of dependent variables. The method is explained below.

Suppose we have a non – linear equation 𝐹(𝑥, 𝑦, 𝑧, 𝑝, 𝑞) = 0 … (0).

We assume that there is an integral of this non – linear equation of the form 𝑢(𝑥, 𝑦, 𝑧) = 0 … (1).

40
Differentiating (1) partially with respect to both x and y gives

𝜕𝑢 𝜕𝑢 𝜕𝑧 𝑢𝑥
+ = 0 𝑜𝑟 𝑢𝑥 + 𝑝𝑢𝑧 = 0 𝑖. 𝑒. 𝑝 = −
𝜕𝑥 𝜕𝑧 𝜕𝑥 𝑢𝑧

𝜕𝑢 𝜕𝑢 𝜕𝑧 𝑢𝑦
+ = 0 𝑜𝑟 𝑢𝑦 + 𝑞𝑢𝑧 = 0 𝑖. 𝑒. 𝑞 = −
𝜕𝑦 𝜕𝑧 𝜕𝑦 𝑢𝑧

We can write
𝑢1 𝑢2
𝑝=− , 𝑞 = − … (2)
𝑢3 𝑢3

𝜕𝑢
𝑤ℎ𝑒𝑟𝑒 𝑢𝑖 = , (𝑖 = 1,2,3)𝑎𝑛𝑑 (𝑥1 , 𝑥2 , 𝑥3 ) = (𝑥, 𝑦, 𝑧).
𝜕𝑥𝑖

Substituting the values of p and q from equation (2) to equation (0) we can obtain a relation of
the form 𝑓(𝑥, 𝑦, 𝑧, 𝑢1 , 𝑢2 , 𝑢3 ) = 0 … (3)

We introduce another two first – order partial differential equations involving arbitrary constants

𝑔(𝑥, 𝑦, 𝑧, 𝑢1 , 𝑢2 , 𝑢3 , 𝑎) = 0, ℎ(𝑥, 𝑦, 𝑧, 𝑢1 , 𝑢2 , 𝑢3 , 𝑏) = 0 … (4)

such that (i) equations (3) and (4) are solvable for 𝑢1 , 𝑢2 , 𝑢3 and

(ii) the equation 𝑑𝑢 = 𝑢1 𝑑𝑥 + 𝑢2 𝑑𝑦 + 𝑢3 𝑑𝑧 … (5) is integrable.

Under these conditions, we can find the solution of the given equation.

Now, equations (3) and (4) must be compatible. Subjecting these equations to the compatibility
condition gives the relation

𝜕𝑔 𝜕𝑔 𝜕𝑔 𝜕𝑔 𝜕𝑔 𝜕𝑔
𝑓𝑢1 + 𝑓𝑢2 + 𝑓𝑢3 − 𝑓𝑥 − 𝑓𝑦 − 𝑓𝑧 =0
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑢1 𝜕𝑢2 𝜕𝑢3

whose auxiliary equations are

𝑑𝑥 𝑑𝑦 𝑑𝑧 𝑑𝑢1 𝑑𝑢2 𝑑𝑢3

= = = = = … (6)
𝑓𝑢1 𝑓𝑢2 𝑓𝑢3 −𝑓𝑥 −𝑓𝑦 −𝑓𝑧

Equations (6) are called Jacobi’s equations. Solving these equations will lead to the desired
solution.

Example

Solve the equation 𝑧 2 = 𝑝𝑞𝑥𝑦 using Jacobi’s method.

41
 𝑢 𝑢
Let 𝐹(𝑥, 𝑦, 𝑧, 𝑝, 𝑞) = 𝑧 2 − 𝑝𝑞𝑥𝑦 = 0 𝑎𝑛𝑑 𝑝 = − 𝑢1 , 𝑞 = − 𝑢2 . Thus,
3 3

𝑢1 𝑢2
𝑧2 − 𝑥𝑦 = 0 𝑜𝑟 𝑢3 2 𝑧 2 − 𝑢1 𝑢2 𝑥𝑦 = 0
𝑢3 𝑢3

𝑖. 𝑒. 𝑓(𝑥, 𝑦, 𝑧, 𝑢1 , 𝑢2 , 𝑢3 ) = 𝑢1 𝑢2 𝑥𝑦 − 𝑢3 2 𝑧 2 = 0.

𝑑𝑥 𝑑𝑦 𝑑𝑧 𝑑𝑢1 𝑑𝑢2 𝑑𝑢3

= = 2
= = =
𝑢2 𝑥𝑦 𝑢1 𝑥𝑦 −2𝑢3 𝑧 −𝑢1 𝑢2 𝑦 −𝑢1 𝑢2 𝑥 2𝑢3 2 𝑧

𝑑𝑥 𝑑𝑢1
=− 𝑜𝑟 𝑢1 𝑥 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 = 𝑎, 𝑠𝑎𝑦,
𝑥 𝑢1

𝑑𝑦 𝑑𝑢2
=− 𝑜𝑟 𝑢2 𝑦 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 = 𝑏, 𝑠𝑎𝑦.
𝑦 𝑢2

𝐻𝑒𝑛𝑐𝑒, 𝑓(𝑥, 𝑦, 𝑧, 𝑢1 , 𝑢2 , 𝑢3 ) = 𝑢1 𝑢2 𝑥𝑦 − 𝑢3 2 𝑧 2 = 𝑎𝑏 − 𝑢3 2 𝑧 2 = 0

√𝑎𝑏
𝑜𝑟 𝑢3 =
𝑧
We now substitute the values of 𝑢1 , 𝑢2 , 𝑢3 in 𝑑𝑢 = 𝑢1 𝑑𝑥 + 𝑢2 𝑑𝑦 + 𝑢3 𝑑𝑧. Hence,

𝑎 𝑏 √𝑎𝑏
𝑑𝑢 = 𝑑𝑥 + 𝑑𝑦 + 𝑑𝑧
𝑥 𝑦 𝑧

Integrating,𝑢 = 𝑎 ln 𝑥 + 𝑏 ln 𝑦 + √𝑎𝑏 ln 𝑧 + 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑜𝑟 𝑢 = ln 𝑥 𝑎 + ln 𝑦 𝑏 + ln 𝑧 √𝑎𝑏 + ln 𝑐

𝑖. 𝑒. 𝑢 = ln 𝑥 𝑎 𝑦 𝑏 𝑧 √𝑎𝑏 + ln 𝑐 𝑜𝑟 𝑢 = ln(𝑐𝑥 𝑎 𝑦 𝑏 𝑧 √𝑎𝑏 ) , 𝑤ℎ𝑒𝑟𝑒 𝑐 𝑖𝑠 𝑎 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡.

1 𝑎 𝑏 1 1
𝑢 = 0 𝑔𝑖𝑣𝑒𝑠 ln(𝑐𝑥 𝑎 𝑦 𝑏 𝑧 √𝑎𝑏 ) = 0 𝑜𝑟 𝑐𝑥 𝑎 𝑦 𝑏 𝑧 √𝑎𝑏 = 1 𝑜𝑟 𝑐 𝑥 𝑦
√𝑎𝑏 √𝑎𝑏 √𝑎𝑏 √𝑎𝑏
[𝑧 ] √𝑎𝑏 = 1√𝑎𝑏 =1

1 √𝑎 √𝑏 1 𝑎
√ √𝑏
𝑖. 𝑒. 𝑐 √𝑎𝑏 𝑥 √𝑏 𝑦 √𝑎 𝑧 = 1 𝑜𝑟 𝑐 √𝑎𝑏 𝑥 𝑏𝑦 𝑎𝑧 =1
1 1
1 𝑎
Thus, 𝑧 = 𝛽𝑥 𝛼 𝑦 𝛼 , 𝑐 √𝑎𝑏 = 𝛽 𝑎𝑛𝑑 √𝑏 = 𝛼.

1
1
Hence 𝑧 = 𝛽𝑥 𝛼 𝑦 𝛼 , 𝛼 𝑎𝑛𝑑 𝛽 𝑏𝑒𝑖𝑛𝑔 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡𝑠.

42