Chapter-10

Classification of Linear Partial Differential

Equations of Order Two and Canonical
Forms

10.1 Introduction
Many problems arising in physical sciences and engineering
when formulated mathematically, give rise to partial differential
equations. Recall that an equation which involves several independent
variables (usually denoted by x , y , z , t ,…), a dependent function u
depending on these independent variables and the partial derivatives
of the dependent function u w.r.t. the independent variables such as
F ( x , y , z ,t , … … ,u , u x , u y , uz , ut , … , u xx ,u yy , u zz ,… , ) =0 …(1)

is called a partial differential equation. A few well-known examples

of partial differential equations which frequently occur in many
practical problems are:

(i) 1-D wave equation: utt =c 2 u xx ,

2
(ii) 2-D wave equation: utt =c ( u xx +u yy ),
2
(iii) 3-D wave equation: utt =c ( u xx +u yy +u zz ) =0,
(iv) 2-D Laplace’s equation: u xx +u yy =0,
(v) 3-D Laplace’s equation: u xx +u yy +u zz =0 ,
(vi) 1-D diffusion equation: ut =k u xx,
(vii) 2-D diffusion equation:ut =k ( u xx +u yy ) ,
 301

(viii) 3-D diffusion equation: ut =k ( u xx +u yy +u zz )

In all the above examples, u is the dependent function and the

suffixes denote partial differentiation with respect to the variables.
Since the order of the partial differential equation is the order of the
highest partial derivative occurring in the equation. Thus, all the
above partial differential equations (i) – (viii) are partial differential
equations of order two.
10.2 Classification of Partial Differential Equations of Order Two
The classification of a partial differential equation depends only
on the highest order partial derivatives present in the equation. This
classification of partial differential equation is motivated by the
classification of the quadratic equation of the form

A x 2+ Bxy +C y 2 + Dx+ Ey + F=0 …(1)

which is elliptic, parabolic or hyperbolic according as the

discriminant B2 – 4AC is negative, zero or positive. Thus, we take a
linear partial differential equation of order two in two independent
variables x and y and a dependent function u depending on x and y in
the form
A u xx + B u xy +C u yy + D u x + Eu y + Fu=G …(2)

where the coefficients A,B,C,… are the functions of x and y;

The sum of first three term of equation (2) determines the type
of partial differential equation and this sum of three terms is called
the principal part of the equation which involves the highest order
derivatives of u. The partial differential equation (2) is elliptic,
parabolic or hyperbolic at a point ( x 0 , y 0 ) according as the
discriminant B2 – 4AC is negative, zero or positive, where
A ≡ A ( x 0 , y 0 ) , B ≡ B(x 0 , y 0 ) and C ≡ C ( x 0 , y 0 ) . If the nature of
equation (3) is same at all points in the given region, then equation (2)
302

is said to be elliptic, parabolic or hyperbolic in that region according

as the discriminant is negative, zero or positive.

SOLVED EXAMPLES

Example 1. Classify the following partial differential equations:

(a) u xx +u yy =0

(b) u xx +4 u xt +4 u tt =0

(c) u xx−4 u xt + 4 utt =0

(d) u xx +uxt +utt =0

(e) 4 uxx +5 u xy +u yy +u x +u y =2 .

Solution. (a) Comparing the given equation with the equation

A u xx + B u xy +C u yy + D u x + Eu y + Fu=G,we have

A=1, B=0 ,C=1 , D=0 , E=0 , F=0 and G=0.

∴ The discriminant ¿ B2−4 AC =−4< 0.

Hence, the given partial differential equation is elliptic.

(b) Comparing the given equation with the equation

A u xx+ B u xt +C utt + Du x + E ut + Fu=G, we have

A=1, B=4, C=4, D=0 , E=0, F=0 and G=0

∴ The discriminant ¿ B2−4 AC =0

Hence, the given partial differential equation is parabolic.

 303

(c) Comparing the given equation with the equation

A u xx+ B u xt +C utt + Du x + E ut + Fu=G, we have

A=1, B=−4 , C=4 , D=0 , E=0 , F=0 and G=0.

∴ The discriminant ¿ B2−4 AC =0

Hence, the given partial differential equation is parabolic.

(d) Comparing the given equation with the equation

A u xx+ B u xt +C utt + Du x + E ut + Fu=G, we have

A=1, B=1 , C=1 , D=0 , E=0 , F=0 and G=0.

∴ The discriminant ¿ B2−4 AC =−3< 0

Hence, the given partial differential equation is elliptic.

(e) Comparing the given equation with the equation

A u xx + B u xy +C u yy + D u x + Eu y + Fu=G ,we have

A=4 , B=5 , C=1 , D=1 , E=1 , F=0 and G=2.

∴ The discriminant ¿ B2−4 AC =21>0

Hence, the given partial differential equation is hyperbolic.

Example 2. Find the region in the xy-plane in which the partial

differential equation is y uxx + ( x + y ) u xy + x u yy =0 parabolic.

Solution. Comparing the given partial differential equation with

A u xx + B u xy +C u yy + D u x + Eu y + Fu=G, we have
304

A= y , B=x + y , C=x , D=E=F=G=0

∴ The discriminant ¿ B2−4 AC =(x+ y)2−4 ( y ) (x )

¿ x 2+ y 2+2 xy−4 xy

¿ x 2+ y 2−2 xy =( x− y)2

The condition for the partial differential equation to be parabolic

2
is B −4 AC =0 . Thus, the region in which the given partial differential
equation is parabolic, is given by

( x− y )2=0 or x− y =0 or x= y

∴ The given partial differential equation is parabolic for the

straight line y=x in the xy – plane.

Example 3. Classify the following partial differential equation:

u xx + x u yy=0, x ≠ 0.

Solution. Comparing the given partial differential equation with

A u xx + B u xy +C u yy + D u x + Eu y + Fu=G , we have

A=1, B=0 ,C=x , D=E=F=G=0.

∴ The discriminant ¿ B2−4 AC =0−4 ( 1 )( x )=−4 x

Hence, the given partial differential equation is of mixed type.

Since it is given that x ≠ 0, therefore, the value of the discriminant
B2−4 AC <0 for all x >0, implying that the given equation is elliptic
for all x >0 and the value of the discriminant B2−4 AC >0 for all x <0,
implying that the given equation is hyperbolic for all x <0.

Example 4. Classify the following partial differential equation

 305

( sin 2 x ) u xx + ( sin 2 x ) u xy + ( cos 2 x ) u yy =x.

Solution. Comparing the given partial differential equation with

A u xx + B u xy +C u yy + D u x + Eu y + Fu=G, we have

A=sin 2 x , B=sin 2 x , C=cos2 x , D=E=F=0∧G=x .

∴ The discriminant ¿ B2−4 AC =¿ ¿

¿( 2sin x cos x)2−(2 sin x cos x)2 =0

Hence, the given partial differential equation is parabolic type.

EXERCISE 10 (A)

1. Classify the following partial differential equations:

(a) 4 uxx +5 u xy +u yy +u x +u y =2

(b) x 2 u xx +2 xy u xy + y 2 u yy =0

(c) u xx +2u xy +u yy=0

(d) u xx−x 2 u yy=u

(e) u xx + x2 u yy =0

2. Find the regions in xy-plane, where the partial differential

equation x u xx + y uxy +u yy =0 is (i) hyperbolic (ii), parabolic and
(iii) elliptic:

ANSWERS
1. (a) hyperbolic (b) parabolic (c) parabolic (d) elliptic (e) elliptic.
306

2. (i) y 2 >4 x (ii) y 2=4 x (iii) y2 < 4x.

10.3 Canonical Forms or Normal Forms

The idea of reducing the given partial differential equation to a
canonical form is that the transformed equation assumes a simple
form so that the subsequent analysis of solving the partial differential
equation is made easy. If the number of independent variables in a
partial differential equation is two or three, a transformation can
always be found so that the given equation is reduced to a canonical
form or normal form. In general, when the number of independent
variables is greater than three, it is not always possible to find such
transformation except in certain special cases.
Let us consider the most general transformation of the
independent variables x and y occurring in the partial differential
equation
A u xx + B u xy +C u yy + D u x + Eu y + Fu=G …(1)

to new independent variables ξ and η, where ξ and η are given by

ξ=ξ(x , y) and η=η(x , y ) …(2)

such that ξ and η are differentiable functions and the Jacobian

∂(ξ ,η) ξ x ξ y
J= =
| |
∂( x , y) ηx η y
=ξ x η y −ξ y ηx ≠0 …(3)

in the region, where equation (1) holds good.

Now, we may write the dependent function u as follows:

u=u [ ξ ( x , y ) , η ( x , y ) ] i.e. u=u(ξ , η) …(4)

Thus, we may find the following

 307

∂u ∂ ξ ∂ u ∂ η
u x= + i .e .u x =u ξ ξ x +uη η x
∂ξ ∂x ∂η ∂ x
∂ u ∂ ξ ∂u ∂ η
uy = + i . e . u y =u ξ ξ y +uη η y
∂ξ ∂ y ∂η ∂ y

u xx =uξξ ξ2x + 2uξη ξ η ηx +uηη η2x +uξ ξ xx +uη η xx …(5)

u xy=u ξξ ξ x ξ y +uξη ( ξ x η y +ξ y η x ) +u ηη η x η y +u ξ ξ xy +u η η xy

and u yy =uξξ ξ 2y +2 uξη ξ y η y +u ηη η2y +u ξ ξ yy +uη η yy

Substituting the values from (4) and (5) in (1), we get

A uξξ + B uξη +C u ηη + D uξ + E uη + F u=G …(6)

where A=A ξ 2x +B ξ x ξ y +C ξ 2y

B=2 A ξ η ηx +B ( ξ x η y + ξ y η x ) +2C ξ y η y

C= A η2x + B η x η y +C η2y …(7)

D= A ξ xx +B ξ xy +C ξ yy + D ξ x + E ξ y

E=A ηxx + B η xy + C η yy + D η x + E η y

F=F∧G=G

It may be noted that the transformed equation (6) has the same form
as the original equation (1) under the general transformation given by (2).

Since the classification of partial differential equation (1)

depends on the coefficients A, B and C, therefore, we can also,
rewrite equation (1) in the following form:

A u xx+ B u xy +C u yy =H ( x , y ,u , u x , u y ) …(8)
308

It can be shown easily that under the transformation (2),

equation (8) takes one of the following three canonical forms:

(i) uξξ −uηη =ϕ ( ξ , η ,u ,uξ ,u η ) or uξη =ϕ (ξ , η ,u , uξ ,u η) in the

hyperbolic case.

(ii) uξξ =ϕ ( ξ , η ,u ,u ξ , uη) or uηη=ϕ (ξ , η , u ,u ξ , uη ) in the parabolic

case.

(iii) uξξ + uηη=ϕ ( ξ , η ,u , uξ ,u η) in the elliptic case.

Now, using the expressions of equation (7), it can be seen that

2
B2−4 A C=( ξ n η y −ξ y ηx ) ( B2−4 AC ) …(9)

and therefore, we conclude that the transformation of the independent

variables does not modify the type of the partial differential equation.

Now, we discuss in detail, the canonical forms for three types of

partial differential equations as given below:

10.4 Canonical Form (Normal Form) for Hyperbolic Partial

Differential Equations

For the hyperbolic partial differential equation, the discriminant

B2−4 A C>0 can be true only when A=0 and C=0. Suppose we set
A=0 and C=0 in equations of (7), which will give us the coordinates
ξ and η that reduce the given partial differential equation to a
canonical form in which the coefficients of uξξ and uηηare zero. Thus,
we have

A=A ξ 2x +B ξ x ξ y +C ξ 2y =0 and C= A η2x + B η x η y +C η2y =0

which can be re-written as

 309

ξn 2 ξx ηx 2 ηx
A ( ) ( )
ξy
+B
ξy
+C=0 and A
ηy
+B ( ) ( )
ηy
+ C=0

Solving these equations for ξ x /ξ y and η x /η y , we get

ξ x −B+ √ B 2−4 AC η x −B−√ B 2−4 AC

= and =
ξy 2A ηy 2A
…(10)

The condition B2−4 AC >0 implies that the slopes of the curves
given by ξ ( x , y )=c 1 and η ( x , y ) =c 2 are real. Thus, if B2−4 AC >0,
then at any point ( x , y ), there exists two real directions given by the
two roots as given in equation (10) along which the partial differential
equation (1) reduces to the canonical form.

Though there are two solutions for each quadratic, we have

considered only one solution for each. Otherwise, we will end up with
the same two co-ordinates.

Now, along the curve ξ ( x , y )=c 1, we have

dξy
dξξ=ξ x dξx +ξ y dξy=0 , which gives =−ξ x /ξ y …(11)
dξx

Again, along the curve η ( x , y ) =c 2, we have

dξy
dξη=η x dξx +η y dξy=0 which gives =−η x /η y …(12)
dξx

These equations (11) and (12) are called characteristic

equations corresponding to the equation (1).

Integrating equations (11) and (12), we obtain the equations of

family of characteristics ξ ( x , y )=c 1 and η ( x , y ) =c 2 which are called
the characteristics of the partial differential equation (1). Now, to
310

obtain the canonical form for the given partial differential equation,
we substitute the expressions of ξ and η into equation (6) which
reduces the equation into the form

uξξ −uηη=ϕ ( ξ , η ,u ,uξ ,u n ) or uξη =ϕ ( ξ , η , u ,u ξ , uη )

To illustrate the procedure and to make the concept more

clear let us consider the following example :

Transform the following PDE into a canonical form:

3 u xx+ 10u xy +3 u yy =0

Solution. Comparing the given partial differential equation with the

equation A u xx + B u xy +C u yy + D u x + Eu y + Fu=G , we get

A=3, B=10,C=3, D=0, E=0, F=0 and G=0.

The discriminant B2−4 AC =100−36=64>0 . Hence, the given

partial differential equation is hyperbolic. The corresponding
characteristic equations are given by

dξy −B+ √ B2−4 AC −10+ 8 1

dξx
=−ξ x /ξ y =− ( 2A
=−
6
=
3 ) ( )
dξy −B−√ B2 −4 AC
and
dξx
=−η x /η y =− (2A
=−
−10−8
6
=3 ) ( )
To find ξ and η, we solve the above two equations for y. Thus,
we get

1
y= x +c 1 and y=3 x +c 2
3

which give the values of the two constants c1 and c2 as given below:
 311

1
c 1= y − x and c 2= y −3 x
3

1
∴ ξ (x , y )≡ y− x=c1 and η(x , y) ≡ y −3 x =c 2
3

These are the equations of straight lines known as characteristic

lines for the given hyperbolic equation. Thus, in this example, the
characteristics are found to be straight lines in xy plane.

From the above characteristic equations, we get

−1
ξ x= , ξ y =1,ξ xx=0, ξ xy =0, ξ yy =0, η x =−3, η y =1,η xx =0,η xy =0 and
3
η yy=0.

To find the canonical form of the given partial differential

equation, we substitute the values of A , B , C ,… and the values of
various partial derivatives ξ x , ξ y , η x, η y,... in the following expressions
to have their values:

A=A ξ 2x +B ξ x ξ y +C ξ 2y =0,

B=2 A ξ x ηx + B ( ξ x η y +ξ y ηx ) +2 C ξ y η y =−64 /3,

C= A η2x + B η x η y +C η2y =0,

D= A ξ xx +B ξ xy +C ξ yy + D ξ x + E ξ y =0

E=A ηxx + B η xy + C η yy + D η x + E η y =0

F=F=0 and G=G=0

∴ Canonical form of the partial differential equation is given by

A uξξ + B uξη +C ξη +C u ηη + D uξ + E uη + F u=G

312

−64
or u =0 or uξη =0
3 ξη

10.5 Canonical Form (Normal From) for Parabolic Partial

Differential Equations

For the parabolic partial differential equation, the discriminant

2
B −4 A C=0 can be true only when B=0 and A or C is equal to zero.
Suppose we set first A=0 in equation (7). Then, we have

A=A ξ 2x +B ξ x ξ y +C ξ 2y =0

ξx 2 ξx
which can be re-written as A ( ) ( )
ξy
+B
ξy
+C=0

ξ x −B ± √ B2−4 AC
Solving this equation for ξ x /ξ y , we get =
ξy 2A

Now, using the condition for parabolic case, we get

ξ x −B
= …(13)
ξy 2 A

Along the curve ξ ( x , y )=c 1, we have

dξy −ξ x
dξξ=ξ x dξx +ξ y dξy=0 , which gives = …(14)
dξx ξ y

From (13) and (14), the characteristic equation is given by

dξy −ξ x B
= = which on integration gives ξ ( x , y )=c 1
dξx ξ y 2 A

In fact, we can verify that A=0 implies B=0 as follows :

 313

B=2 A ξ x ηx + B ( ξ x η y +ξ y ηx ) +2 C ξ y η y

Since B2−4 A C=0 which gives B=2 √ AC and so, we have

B́=2 A ξ x ηx +2 √ AC ( ξ x η y + ξ y η x ) + 2C ξ y η y

or B=2 ( √ A ξ x + √ C ξ y )( √ A η x + √ C η y )

ξ x −B −2 √ AC −√ C
Also, we have = = = which gives
ξy 2 A 2A √A
√ C ξ y =− √ A ξ x
∴ B=2 ( √ A ξ x − √ A ξ x )( √ A η x + √ C η y ) =2 ( 0 ) ( √ A η x + √C η y ) =0

We, therefore, choose ξ in such a way that both A and B are

zero. Then, η can be chosen in any way, we like as long as it is not
parallel to the ξ- coordinate. In other words, we choose η such that the
Jacobian of the transformation is not zero. Thus, we can write the
canonical form for the parabolic partial differential equation by simply
substituting ξ and η into equation

A uξξ + B uξη +C u ηη + D uξ + E uη + F u=G

which reduces to either of the following form

uξξ =ϕ ( ξ , η, u , uξ ,uη ) or uηη =ϕ ( ξ , η ,u , uξ ,u η ) .

To illustrate the procedure, let us consider the following example;

Transform the following PDE into a canonical form:

x 2 u xx −2 xy uxy + y 2 u yy =e x

Solution. Comparing the given partial differential equation with

314

A u xx + B u xy +C u yy + D u x + Eu y + Fu=G , we get

A=x2 , B=−2 xy ,C= y 2 , D=0, E=0, F=0 and G=e x

∴ The discriminant ¿ B2−4 AC =4 x 2 y 2−4 x 2 y 2=0

Hence, the given partial differential equation is parabolic every

where. The corresponding characteristic equation is given by

dξy −ξ x B −2 xy − y dξy −dξx

= = = = or =
dξx ξ y 2 A 2x2
x y x

which on integration gives log y=−log x+ log c or xy=c

Hence, ξ=xy will satisfy the characteristic equation and we can

choose η= y. Now, to find the canonical form of the given partial
differential equation, we substitute the expressions for ξ and η into
equations (7) to get

A=A y 2+ Bxy +C x 2=x 2 y 2−2 x 2 y 2 + y 2 x2=0,

B=0, C= y 2 , D=−2 xy , E=0, F=0,∧G=e x

Hence, the transformed equation of the given PDE is

y 2 u ηη−2 xy uξ =e x or η2 uηη =2 ξ u ξ +e ξ/ η

∴ The required canonical form is given by uηη= ( 2ξη ) u +( η1 ) e

2 ξ 2
ξ/ η

10.6 Canonical Form for Elliptic Partial Differential Equations

For the elliptic partial differential equation, the discriminant

B2−4 A C<0.
 315

Doing the similar analysis as in case of canonical form for

hyperbolic partial differential equation, the characteristic equations are
given by

dξy B−√ B2−4 AC dξy B+ √ B2−4 AC

= and =
dξx 2A dξx 2A

which give us complex conjugate co-ordinates, say ξ and η.

Now, we make another transformation from coordinates (ξ , η) to

the new co-ordinates(α , β ) given by

1 1
α = ( ξ +η ) and β= (ξ−η)
2 2i

which provide us the desired canonical form of the equation as

uαα +u ββ =ϕ ( α , β , u ,u α , u β ).

To illustrate the procedure, let us consider the following example:

Transform the following PDE into a canonical form

u xx + x2 u yy =0, ( x >0 ) .

Solution. Comparing the given partial differential equation

A u xx + B u xy +C u yy + D u x + Eu y + Fu=G , we get

A=1, B=0,C=x 2 , D=0 , E=0 , F=0 and G = 0

∴ The discriminant ¿ B2−4 AC =0−4 x2=−4 x 2 <0, x> 0.

Hence, the given partial differential equation is elliptic.

∴ The characteristic equations are given by

316

dξy B−√ B2−4 AC −√ −4 x 2

= = =−ix
dξx 2A 2

dξy B+ √ B2−4 AC √−4 x 2

and = = =ix
dξx 2A 2

Integrating these two equations, we get

x2 x2
iy+ =c 1 and −iy+ =c 2
2 2

x2 x2
Now, we assume that ξ= +iy and η= −iy
2 2

1
Again, introducing the transformations α = (ξ+ η) and
2
1
β= (ξ−η), we obtain
2i

1 x2 x2 x2
α= (
2 2
+iy + −iy =
2 2 )
1 x2 x2
and β= (
2i 2
+iy− +iy = y
2 )
∴ The transformed equation is obtained by first computing

A=A α 2x + B α x α y + C α 2y =x 2,

B=3 A α x β x +B ( α x β y +α y β x ) + 2C α y β y =0,

2 2 2
C= A β x + B β x β y +C β y =x ,

D= A α xx + B α xy+C α yy + D α x + E α y =1 ,
 317

E=A β xx +B β xy + C β yy + D β x + E β y =0,

F=F=0 and G=G=0.

and then substituting in the following equation

A uαα + Bu αβ +C u ββ + D uα + E u β + F u=G

Thus, canonical form of the given partial differential equation is

x 2 u αα + x 2 u ββ +u α =0

−uα x2
or uαα +u ββ = , since α =
2α 2

The concept of reduction of partial differential equations in

canonical form is made more clear by following examples:

SOLVED EXAMPLES

Example 1. Reduce the following partial differential equation to a

canonical form and hence solve it :

u xx−2 ( sin x ) u xy−( cos2 x ) u yy −( cos x ) u y =0

Solution. Comparing the given partial differential equation with

A u xx + B u xy +C u yy + D u x + Eu y + Fu=G , we have

A=1, B=−2sin x ,C=−cos2 x , D=0 , E=−cos x , F=G=0

∴ The discriminant ¿ B2−4 AC =4 sin 2 x+ 4 cos 2 x=4>0

Hence, the given partial differential equation is hyperbolic.

The corresponding characteristic equations are given by

318

dξy B−√ B2−4 AC −2 sin x −2

= = =−sin x−1
dξx 2A 2
…(1)

dξy B+ √ B2−4 AC −2 sin x−2

and = = =−sin x +1 …
dξx 2A 2
(2)

Integrating equations (1) and (2), we get

y=cos x−x+ c 1 and y=cos x+ x +c 2

Thus, we choose the characteristics as

ξ=x + y−cos x=c 1 and η=−x+ y−cos x=c 2 …(3)

In order the find canonical form of the equation, we compute

A=A ξ 2x +B ξ x ξ y +C ξ 2y =0,

B=2 A ξ x ηx + B ( ξ x η y +ξ y ηx ) +2 C ξ y η y

¿ 2 ( sin x+1 ) ( sin x−1 )−4 sin 2 x−2 cos2 x

¿ 2 ( sin2 x−1 )−4 sin 2 x−2 cos2 x

¿−2−2 ( sin2 x+ cos2 x ) =−4,

C=0, D=0, E=0, F=0∧G=0

Thus, the canonical form of the equation is given by

A uξξ + B uξη +C u ηη + D uξ + E uη + F u=G

or −4 uξη=0 or uξη =0
 319

Integrating uξη =0 w.r.t. ξ, we obtain

uη =f ( η ) , where f (η) is an arbitrary function of η .

Now, integrating uη =f ( η) w.r.t. η, we obtain

u=ηf ( η ) + g(ξ ) …(4)

where g( ξ) is another arbitrary function of ξ.

Finally, substituting the values of ξ and η from (3) in (4), we get

the required solution of the given partial differential equation as

u ( x , y )= ( y−x−cos x ) f ( y−x −cos x )+ g ¿

Example 2. Classify and reduce the following partial differential

equation to a canonical form and solve it :

y 2 u xx −2 xy u xy + x2 u yy −( y 2 / x ) u x −( x 2 / y ) u y =0

Solution. Comparing the given partial differential equation with

A u xx+ B u x y +C u yy + D u x + E u y + Fu=G, we have

A= y 2 , B=−2 xy ,C=x 2 , D=− y 2 / x , E=−x 2 / y , F=G=0 .

∴ The discriminant ¿ B2−4 AC =(−2 xy )2−4 ( y 2 ) ( x 2)

¿ 4 x2 y 2−4 x 2 y 2=0

Hence, the given equation is a parabolic type of PDE.

∴ The characteristic equation is given by

dξy −ξ x B −2 xy
= = = =−x / y
dξx ξ y 2 A 2 y 2
320

which on integration gives x 2+ y 2=c 1

∴ ξ=x 2 + y 2 satisfies the characteristic equation. The η-

coordinate can be chosen arbitrary so that it is not parallel to ξ i.e. the
Jacobian of the transformation is not zero. Thus, we choose

ξ=x 2 + y 2 and η= y 2 …(1)

Now, to find the canonical form of the equation, we compute

A=A ξ 2x +B ξ x ξ y +C ξ 2y =4 x 2 y 2−8 x 2 y 2+ 4 x 2 y 2 =0,

B=0, C=4 x 2 y 2 , D=E=F=G=0

Hence, the canonical form of the equation is given by

A uξξ + B uξη +C u ηη + D uξ + E uη + F u=G

or 4 x2 y 2 u ηη=0 or uηη=0 …(2)

To solve uηη=0, we integrate it twice w.r.t. η to get

uη =f ( ξ) and u=ηf ( ξ )+ g (ξ)

where f (ξ) and g( ξ) are arbitrary functions of ξ.

Finally, substituting the values of ξ and η from (1) in (2), we get

u ( x , y ) = y 2 f ( x 2 + y 2 ) + g ( x 2+ y 2 ) …(3)

Example 3. Reduce the following partial differential equation to a

canonical form :

u xx +2u xy + 4 u yy +2 u x +3 u y =0.

Solution. Comparing the given partial differential equation with

 321

A u xx + B u xy +C u yy + D u x + Eu y + Fu=G , we have

A=1, B=2, C=4 , D=2 , E=3 , F=0∧G=0.

∴ The discriminant ¿ B2−4 AC =−12<0

Hence, the given equation is an elliptic type of PDE.

The characteristic equations are given by

dξy B−√ B2−4 Ac 2(1+i √3)

= = =1−i √ 3
dξx 2A 2
…(1)

dξy B+ √ B2−4 Ac 2(1+ i √ 3)

and = = =1+i √ 3 …
dξx 2A 2
(2)

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

y= ( 1−i √3 ) x +c 1 and y= ( 1+i √ 3 ) x+ c2

Hence, we may take the characteristics as

ξ= y −( 1−i √ 3 ) x and η= y−( 1+i √3 ) x …(3)

Now, in order to avoid calculations with complex variables, we

introduce the second transformation given by

1 1
α = ( ξ +η ) = y−x and β= ( ξ−η )=√ 3 x …(4)
2 2λ

The canonical form can now be obtained by computing

A=A α 2x + B α x α y + C α 2y =3,
322

B=2 A α x β x + B ( α x β y + α y β x ) +2 C α y β y =0,

C= A β 2x + B β x β y +C β 2y =3,

D= A α xx + B α xy +C α yy + D α x + E α y =1 ,

E=A β xx +B β xy + C β yy + D β x + E β y =2 √3 ,

F=0∧G=0

Thus, the required canonical form is given by

A uαα + Bu αβ +C u ββ + D uα + E u β + F u=G

or 3 uαα +3 u ββ + uα + 2 √ 3 u β=0

−1
or uαα +u ββ= ( u + 2√ 3 uβ) …(5)
3 α

Example 4. Show that the partial differential equation

( 2xN )u =( a1 ) u ,where N and a are constants, is

u xx + x 2 yy

hyperbolic. Obtain its characteristic equations.

Solution. Comparing the given partial differential equation with

A u xx + B u xy +C u yy + D u x + Eu y + Fu=G, we have

2N
A=1, B=0 ,C=−1 / a2 , D= , E=0 , F=0 ,G=0.
x

∴ The discriminant ¿ B2−4 AC =0−4 ( 1 ) (−1 /a2 )=4/ a2> 0

Hence, the given partial differential equation is hyperbolic.

 323

∴ The characteristic equations are given by

dξy B−√ B2−4 Ac 0−√ 4 /a2

= = =−1/a …
dξx 2A 2
(1)

dξy B+ √ B2−4 Ac 0+ √ 4 /a2

and = = =1/a …
dξx 2A 2
(2)

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

−x x
y= + c 1 and y= + c2
a a

or x +ay=c1 and x−ay =c 2

∴ The characteristic equations are given by

ξ=x +ayand η=x−ay …(3)

EXERCISE 10 (B)

1. Find the characteristic equations of the partial differential equation

u xx +2u xy + ( sin 2 x ) u yy +u y =0, when it is of hyperbolic type.

2. Reduce the Tricomi equation u xx + x u yy =0, ( x ≠ 0 ) to canonical

forms.

3. Reduce 3 u xx+ 10u xy +3 u yy =0 to a canonical form and hence solve

it.

4. Find the families of characteristics of the partial differential

equation ( 1−x 2 ) u xx −u yy =0 in the elliptic and hyperbolic cases.
324

5. Find the characteristics of the following partial differential

equation u xx +2u xy + ( cos 2 u ) u yy +2u x +3 u y =0, when it is of
hyperbolic type.

6. Transform the following partial differential equation to a canonical

form:

( sin 2 x ) u xx + ( sin 2 x ) u xy + ( cos 2 x ) u yy =x.

7. Classify and reduce the following partial differential equation to a
canonical form

( 1+ x 2 ) u xx + ( 1+ y 2 ) u yy + x u x + y u y =0.
8. Classify and reduce the following partial differential equations to
canonical form y 2 u xx + x 2 u yy=0

ANSWERS
1. ξ= y −x+ sin x and η= y−x−sin x .
1
2. uαα +u ββ + u =0. 3. uξn =0
3β β
3 3
4. Hyperbolic for x < 0: Characteristic Equations: ξ= y −( √ −x ) ,
2
3 3
η= y + ( √−x ) ,Canonical Form: 6 ( ξ−η ) u ξη=uξ −u η.
2
3 3
Elliptic for x > 0 : Characteristic Equations: ξ= y −i ( √ x ) ,
2
3 3
η= y +i ( √ x ) , Canonical Form: uαα +u ββ + ( 1/3 β ) u β=0.
2
6. Parabolic PDE, Characteristic Equations: ξ= y −ln sin y ,η= y,
Canonical Form: [ 1−e2 (η−ξ) ] uηη =sin−1 ( eη−ξ )−u ξ.
 325

7. Elliptic PDE, Characteristics Equations:

2 2
ξ=ln ( x+ √ x +1 )−i ln ( y + √ y +1 ),
η=ln ( x + √ x2 +1 ) +i ln ( y + √ y 2 +1 ), Canonical Form: uαα +u ββ=0.
8. Elliptic PDE, Characteristic Equations: ξ= y 2 +i x2, η= y 2−i x 2
1 1 1
Canonical Form: uαα +u ββ +
2 α (
u∝ + u β =0 .
β )
OBJECTIVE TYPE QUESTIONS

1. The PDE y 3 uxx −( x 2−1 ) u yy =0 is

(a) Parabolic in { ( x , y ) : x <0 } (b) Hyperbolic in { ( x , y ) : y >0 }

(c) Elliptic in R2 (d) Parabolic in { ( x , y ) : x >0 }

2. The characteristic curves of x 2 u xx − y 2 u yy =x2 y 2 + x , x> 0 are

(a) Rectangular hyperbolas (b) Parabolas

(c) Circles (d) Straight lines

3. The region in which the PDE y uxx +2 xy uxy + x u yy =ux +u y is

hyperbolic is

(a) xy ≠1 (b) xy ≠0

(c) xy >1 (d) xy >0

4. The PDE u xx −2 ( sin x ) u xy ( cos2 x ) u yy is

(a) Parabolic (b) Hyperbolic

(c) Elliptic (d) None of these

326

5. The characteristic curves of u xx + ( 2xN )u =( a1 ) u , where N and a

x 2 yy

are constants, are given by

(a) x +ay=c1 and x−ay =c 2

(b) x + y=c 1 and x− y =c 2

(c) ax + y=c and ax− y=c 2

(d) None of these

ANSWERS

1. (b) 2. (d) 3. (c) 4. (b) 5. (a)