Linear PDE With Constant Coefficients: Pinaki Pal

Linear PDE with Constant Coefficients
Pinaki Pal
Department of Mathematics
National Institute of Technology Durgapur
West Bengal, India
pinaki.pal@maths.nitdgp.ac.in
Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 1 / 54

Linear PDE
A linear PDE of order n in the dependent variable z and independent
variables x and y can be represented in the form
∂nz ∂nz ∂ n z ∂ n−1 z ∂ n−1 z ∂ n−1 z

f( , , . . . , , , , . . . , ,
∂x n ∂x n−1 ∂y ∂y n ∂x n−1 ∂x n−2 ∂y ∂y n−1
∂z ∂z
..., , , z, x, y ) = 0, (1)
∂x ∂y
where the powers of the dependent variable and its derivatives are strictly
equal to one. Moreover, no product of the dependent variable and its
derivatives can appear in any term of the equation (1).

Examples
Check whether the following PDEs are linear or not
∂ z2 ∂z
1 (x 2 + y ) ∂x 2 + 3 ∂y + z = x
3 (Linear PDE of order 2)
3
∂ z ∂z 2
2 ∂y 3 + (3x − y ) ∂y + x z = yx (Linear PDE of order 3)
2
∂4z ∂z
3
∂x 4
− 5 ∂y + 2z = yx (Linear PDE of order 4)
h i2
∂4z ∂z
4
∂x 4
− 5 ∂y + 2z = yx (Not linear)
∂4z ∂z
5
∂x 4
− 5 ∂y + 2z 2 = yx + 3 (Not linear)
∂4z ∂z
6
∂y 4
− 5 ∂y z = yx 2 (Not linear)

Linear PDE with constant coefficients
A linear PDE, in which coefficients of the dependent variable and its
derivatives are constants is called a linear PDE with constant coefficients.
Examples
Check whether the following equations are linear PDE with constant
coefficients
∂4z ∂z
1
∂y 4
− 5 ∂y = yx 2 (Yes)
∂2z ∂2z ∂z
2
∂y 2
+ ∂x∂y − 5 ∂y = y (Yes)
∂3z ∂z
3
∂y 3
− 5 ∂y = 0 (Yes)
∂3z ∂ z 2 ∂z
4
∂y 3
+ 2x ∂x∂y − 5 ∂y = 5x 2 (Linear but not with variable coefficients)
3
∂ z ∂ z 3 ∂z
4 ∂y 3 + 3 ∂x∂y 2 − 5 ∂y + 7z = x + sin y (Yes)
5

Linear PDE with constant coefficients can further be divided into two
different classes namely homogeneous and nonhomogeneous.
Liner homogeneous PDE with constant coefficients
An equation of the form
∂z n ∂z n ∂z n ∂z n ∂z n
+a1 +a 2 +· · ·+a n−1 +an = f (x, y ),
∂x n ∂x n−1 ∂y ∂x n−2 ∂y 2 ∂x∂y n−1 ∂y n
(2)
where a1 , a2 , . . . and an are constants and f is a function of the
independent variables, is called a linear homogeneous PDE of order n
with constant coefficients. Note that all the derivatives appeared in the
equation are of same order.
Liner nonhomogeneous PDE with constant coefficients

A linear PDE with constant coefficients consisting of at least two
derivatives of different orders or a term involving the dependent variable is
called nonhomogeneous linear PDE with constant coefficients.

Examples
∂2z ∂ z2 ∂ z 2
∂x 2
+ 5 ∂x∂y − 3 ∂y 2 = x sin x (Homogeneous)
∂3z 3
∂x 3
+ 7 ∂x∂2 ∂y
z
= 0 (Homogeneous)
∂3z ∂2z
∂x 3
+ ∂x∂y = 0 (Nonhomogeneous)
∂4z ∂4z
∂y 4
+ ∂x 2 ∂y 2
− 3z = 0 (Nonhomogeneous)
∂3z ∂2z
∂x 3
+ ∂x∂y = x (Nonhomogeneous)

General solution of the homogeneous linear PDE with constant
coefficients
Consider the homogeneous linear PDE
∂z n ∂z n ∂z n ∂z n ∂z n
+a1 +a 2 +· · ·+a n−1 +an = f (x, y ),
∂x n ∂x n−1 ∂y ∂x n−2 ∂y 2 ∂x∂y n−1 ∂y n
(3)
where a1 , a2 , . . . and an are constants and f is a function of the
independent variables. Introducing the notations D = ∂x ∂
and D 0 = ∂y∂
we
can write the above equation in the form
F (D, D 0 )z = f (x, y ), (4)
where f (D, D 0 ) = D n + a1 D n−1 D 0 + · · · + an−1 DD 0n−1 + an D 0n . Now to

find the general solution of the PDE (3), we first find the general solution
of it when f (x, y ) = 0. In that case the PDE (3) reduces to
F (D, D 0 )z = 0. (5)

Let z = φ(y + mx) be a solution of the PDE (5). Then substituting z in
(5) we get
(mn + a1 mn−1 + a2 mn−2 + · · · + an−1 m + an )φn (y + mx) = 0 (6)

=⇒ mn + a1 mn−1 + a2 mn−2 + · · · + an−1 m + an = 0. (7)
Now we observe that the equation (7), called the auxiliary equation is a
polynomial of degree n and let us suppose the not necessarily distinct
roots of the polynomial are m1 , m2 , m3 , . . . , mn . In the following we
consider following two possible cases.
Distinct roots: m1 , m2 , m3 , . . . , mn are distinct real or complex
numbers. The complete solution of the PDE (5) is then given by
z = φ1 (y + m1 x) + φ2 (y + m2 x) + · · · + φn (y + mn x),
where φ1 , φ2 , . . . are arbitrary functions.

Repeated roots: Let a particular root m of the equation (7) is
repeated r times. Then the contribution of this root to the complete
solution of the PDE (5) will be
x r −1 φ1 (y + mx) + x r −2 φ2 (y + mx) + · · · + φr (y + mx),
while the contributions of the distinct roots to complete solution will

be same as the previous case. The following examples will illustrate
the procedure. Note that the compete solution of the PDE (5) is
called the complementary function (CF) of the PDE (2).
Examples
Solve the PDEs: (i)(D 3 − 6D 2 D 0 + 11DD 02 − 6D 02 )z = 0,
(ii) (D 3 − 3D 2 D 0 + 3DD 02 − D 03 )z = 0, (iii) (D 4 − D 04 )z = 0,
(iv) (D 4 − 2D 3 D 0 + 2DD 02 − D 04 )z = 0.

Solution (i)
Given equation is (D 3 − 6D 2 D 0 + 11DD 02 − 6D 02 )z = 0.
Auxiliary equation is m3 − 6m2 + 11m − 6 = 0.
The roots of the auxiliary equation are m = 1, 2, 3, which are distinct.
Therefore, complete solutions of the given PDE is
z = φ1 (y + x) + φ2 (y + 2x) + φ3 (y + 3x),
where φ1 , φ2 and φ3 are arbitrary functions.
Solution (ii)
Given equation is (D 3 − 3D 2 D 0 + 3DD 02 − D 03 )z = 0.
The roots of the auxiliary equation are m = 1, 1, 1, which are repeated.
z = x 2 φ1 (y + x) + xφ2 (y + x) + φ3 (y + x),
where φ1 , φ2 and φ3 are arbitrary functions.

Solution (iii)
Given equation is (D 4 − D 04 )z = 0.
Auxiliary equation is m4 − 1 = 0.
The roots of the auxiliary equation are m = 1, −1, i, −i, which are distinct. Note
that there are complex conjugate roots. Therefore, complete solutions of the
given PDE is
z = φ1 (y + x) + φ2 (y − x) + φ3 (y + ix) + φ4 (y − ix),
where φ1 , φ2 , phi3 and φ4 are arbitrary functions.
Solution (iv)
Given equation is (D 4 − 2D 3 D 0 + 2DD 02 − D 04 )z = 0.
The roots of the auxiliary equation are m = 1, 1, 1, −1, which are distinct.
z = x 2 φ1 (y + x) + xφ2 (y + x) + φ3 (y + x) + φ4 (y − x),
where φ1 , φ2 , phi3 and φ4 are arbitrary functions.

Now we consider the the PDE
F (D, D 0 )z = f (x, y ),
again and define the Particular Integral (PI) of as follows.
Particular Integral
It is a function u of x and y denoted by the symbol
1
f (x, y ).
F (D, D 0 )
The function u is such that the action of the operator F (D, D 0 ) on u

returns the function f (x, y ). i.e.
F (D, D 0 )u = f (x, y ).
Clearly u(x, y ) is a particular function which satisfy the PDE homogeneous

PDE (4). Note that PI does not involve any arbitrary function or constant.

General Method
1 1
Meaning of D
f (x, y ) and D0
f (x, y )
Let
1 ∂X
f (x, y ) = X (x, y ), =⇒ = f (x, y )
D Z ∂x
Integrating, X = f (x, y )dx (Partil Integration w.r.t. x)
1
Similarly D 0 f (x, y ) means partial integration of f (x, y ) w.r.t. y .
.
Now let m1 , m2 , . . . , mn are the roots of the auxiliary equation. Then
F (D, D 0 ) = (D − m1 D 0 )(D − m2 D 0 ) . . . (D − mn D 0 ),

which implies
1 1
f (x, y ) = f (x, y )
F (D, D 0 ) (D − m1 D 0 )(D
− m2 D 0 ) . . . (D − mn D 0 )
1
= X1 , (8)
(D − m1 D )(D − m2 D 0 ) . . . (D − mn−1 D 0 )
0
1
where X1 = (D−mn D 0 ) f (x, y ).
Therefore,
(D − mn D 0 )X1 = f (x, y ),
∂X1 ∂X1
Or, − mn = f (x, y ). (9)
∂x ∂y
Now the equation (9) is a Lagrange’s equation whose subsidiary equation

is given by
dx dy dX1
= = . (10)
1 −mn f (x, y )

From the first two terms of the simultaneous differential equations (10) we
get
y = c − mn x,
where c is an arbitrary constant. Now from the first and third terms of
equation (10), we get
Z
X1 = f (x, c − mn x)dx
and after integration we have to replace c by y + mn x.

Therefore, to find the particular integral
1
f (x, y )
F (D, D 0 )
we need to repeat the above procedure n number of times. Although the

method is general but often it is lengthy. In the following we discuss some
short methods for finding particular integral.

Short Methods
I: f (x, y ) = x m y n , m and n are some positive integers

1
In this case the the function F (D,D 0 ) is either expanded in an infinite series
0
of ascending powers of D or D and action of the resulting infinite series
operator on f (x, y ) is determined to find the PI.
Examples
Find particular integrals of the PDEs (i) (D 2 − a2 D 02 )z = x(a = constant),
(ii) (D 2 + 3DD 0 + 2D 02 )z = x + y , (iii) (D 2 − DD 0 − 6D 02 )z = xy

Solution (i)
Here f (x, y ) = x.
1
∴ PI = x
D2− a2 D 02
−1
D2

1
= − 2 02 1 − 2 02 x
a D a D
D2

1
= − 2 02 1 + 2 02 + . . . x
a D a D
1 1 1
= − 2 02 x = − 2 · 02 x
a D a D
xy 2
= − 2.
2a

Solution (ii)
1
PI = (x + y )
D 2 + 3DD 0 + 2D 02
0 −1
2D 02

1 3D
= 1+ + 2 (x + y )
D2 D D
3D 0

1
= 1− + ...
D2 D

1 3
= x + y − 1
D2 D
1 1
= (y + x − 3x) = 2 (y − 2x)
D2 D
x 2y x3
= − .
2 3

Solution (iii)
1
PI = xy
D2 − DD 0 − 6D 02
−1
D 02

1 D
= 1− 0 −6 2 xy
D2 D D

1 D
= 2
1 + 0 + . . . xy
D D
x2

1
= xy +
D2 2
1 3 x 4
= x y+
6 24

Lecture 5: 01.10.2020

II: f (x, y ) = φ(ax + by )
Under this case there are two possibilities.
(a) F (a, b) 6= 0
ZZ
1 1
PI = φ(ax + by ) = . . . n times φ(t)dtdt . . . dt,
F (D, D 0 ) F (a, b)
where n is the order of the PDE and t = ax + by .

(b) F (D, D 0 ) = (bD − aD 0 )r ψ(D, D 0 ) such that ψ(a, b) 6= 0.
xr
ZZ
1 1
PI = 0
φ(ax+by ) = . . . (n−r ) times ψ(t)dtdt . . . dt.
F (D, D ) r !b r ψ(a, b)
Examples
Find the particular integrals for the following PDEs.
(i) (D 2 + 2DD 0 + D 02 )z = e 2x+3y , (ii) (D 2 + 3DD 0 + 2D 02 )z = x + y ,
(iii) (D 3 − 4D 2 D 0 + 4DD 02 )z = cos(2x + y ).
Solution (i)
Here
F (D, D 0 ) = D 2 + 2DD 0 + D 02 ,
f (x, y ) = e 2x+3y is of the form φ(ax + by ) with a = 2 and b = 3 and
F (a, b) = F (2, 3) = 22 + 2 × 2 × 3 + 32 = 25 6= 0.
Therefore,
1
PI = e 2x+3y
D 2 + 2DD 0 + D 02 Z Z
1
= ( e t dt)dt
22 + 2 × 2 × 3 + 32
1 t 1
= e = e 2x+3y
25 25

Solution (ii)
Here F (D, D) = D 2 + 3DD 0 + 2D 02 , f (x, y ) = x + y is of the form
φ(ax + by ) with a = b = 1 and F (1, 1) = 6 6= 0. Therefore,
1
PI = (x + y )
D2 + 3DD 0 + 2D 02 Z Z
1
= ( tdt)dt
12 + 3 × 1 × 1 + 2 × 12
1 t3 (x + y )3
= = .
6 6 36

Solution (iii)
Here F (D, D) = D 3 − 4D 2 D 0 + 4DD 02 , f (x, y ) = cos(2x + y ) is of the
form φ(ax + by ) with a = 2, b = 1 and F (2, 1) = 0.
Now we note that F (D, D 0 ) = D(D − 2D 0 )2 . So the particular integral is
given by
1
PI = cos(2x + y )
D3 − 4D 2 D 0 + 4DD 02
1 x2 1
= cos(2x + y ) = cos(2x + y )
D(D − 2D 0 )2 2!12 D
x2 1 x2
Z
= × cos tdt = sin t
2 2 4
x2
= sin (2x + y ).
4

Complete solution
Consider the homogeneous linear PDE of the form
F (D, D 0 )z = f (x, y ),
with constant coefficients. Then we know how to find the

complementary function (CF) and particular integral (PI) of the PDE.
Now if we know the CF and PI of the PDE, then the complete solution
of the PDE is given by
z = CF + PI .

Examples
Solve the PDEs
∂2z ∂2z
(i)
∂x 2
− ∂x∂y = cos x cos 2y
∂3z ∂3z
(ii)
∂x 3
−2 ∂ 2 x∂y
= 2e 2x + 3x 2 y
∂2z ∂2z ∂ z2
(iii)
∂x 2
+ ∂x∂y − 6 ∂y 2 = y cos x

Solution (i)
Auxiliary Equation is
m2 − m = 0 and the roots of it are m = 0, 1.
Therefore complementary function is z = φ1 (y ) + φ2 (y + x).

Now
1 1 1
PI = 0
cos x cos 2y = [cos(x + 2y ) + cos(x − 2y )]
D2 − DD 2 D − DD 0
2
ZZ ZZ
1 1 1
= cos tdtdt + cos t1 dt1 dt1
2 1−2 1+2

1 1
= cos t − cos t1
2 3
1 1
= cos(x + 2y ) − cos(x − 2y )
2 6
Hence the complete solution is give by,
1 1
z = φ1 (y ) + φ2 (y + x) + cos(x + 2y ) − cos(x − 2y ).
2 6
Solution (ii)
Auxiliary Equation is m3 − 2m2 = 0 and the roots of it are m = 0, 0, 2.
Therefore complementary function is z = xφ1 (y ) + φ2 (y ) + φ3 (y + 2x).
Now
1 1 1
PI = (2e 2 x + 3x 2 y ) = 2 3 e 2x + 3 3 x 2y
D 3 − D 2D 0 D − D 2D 0 D − D 2D 0
1 3 1
= 2 3 e 2x + 3 2
0 x y
2
2 −2 ×0 D 1 − DD
−1
D0

1 2x 3
= e + 3 1− x 2y
4 D D
D0 D 02

1 2x 3
= e + 3 1+ + 2 + . . . x 2y
4 D D D
x3

1 2x 3 1 1 3
= e + 3 x 2 y + x 2 = e 2x + 3 x 2 y +
4 D D 4 D 3
1 2x x 5 y x6
= e + +
4 20 60

So the complete solution is given by
1 x 5y x6
z = xφ1 (y ) + φ2 (y ) + φ3 (y + 2x + e 2x + + .
4 20 60
Solution (iii)
Auxiliary Equation is m2 + m − 6 = 0 and the roots of it are m = 2, −3.
Therefore complementary function is z = φ1 (y + 2x) + φ2 (y − 3x).
Now
1 1
PI = 0 02
y cos x = y cos x
D2 + DD − 6D (D + 3D )(D − 2D 0 )
0
Z
1
= (c − 2x) cos xdx [∵ y = c − 2x]
(D + 3D 0 )
1
= [c sin x − 2x sin x − cos x]
(D + 3D 0 )
1
= [(y + 2x) sin x − 2x sin x − cos x]
(D + 3D 0 )
1
∴ PI = [y sin x − cos x]
(D + 3D 0 )
Z
= [(c + 3x) sin x − cos x] dx [∵ y = c + 3x]
= −c cos x − sin x − 3x cos x + 3 sin x
= −(y − 3x) cos x − sin x − 3x cos x + 3 sin x
= −y cos x + 2 sin x
Hence the complete solution is given by
z = φ1 (y + 2x) + φ2 (y − 3x) − y cos x + 2 sin x.

Lecture 6: 08.10.2020

Classification of Second Order PDE
We consider the most general second order linear PDE of the form
∂2u ∂2u ∂2u ∂u ∂u
A 2
+ B + C 2
+D +E + Fu = G , (11)
∂x ∂x∂y ∂y ∂x ∂y
where A, B, C , D, E , F , G are functions of the independent variables x and
y.
We call the function
B(x, y )2 − 4A(x, y )C (x, y )
as discriminant of the equation (11).

Classification
At a point (x0 , y0 ) in its domain, the PDE (11) is called
Hyperbolic if B(x0 , y0 )2 − 4A(x0 , y0 )C (x0 , y0 ) > 0.
Parabolic if B(x0 , y0 )2 − 4A(x0 , y0 )C (x0 , y0 ) = 0.
Elliptic if B(x0 , y0 )2 − 4A(x0 , y0 )C (x0 , y0 ) < 0.
Cannonical Forms
Consider the most general of the independent variables x and y of the
PDE (11) to new variables ξ and η given by
ξ = ξ(x, y ), η = η(x, y )
such that the functions are continuously differentiable and the Jacobian

∂(ξ, η) ξx ξy
J(x, y ) = = 6= 0,
∂(x, y ) ηx ηy
in the domain of definition of the PDE. Using the chain rule we now have
the following
ux = uξ ξx + uη ηx
uy = uξ ξy + uη ηy
uxx = uξξ ξx2 + 2uξη ξx ηx + uηη ηx2 + uξ ξxx + uη ηxx
uxy = uξξ ξx ξy + uξη (ξx ηy + ξy ηx ) + uηη ηx ηy + uξ ξxy + uη ηxy
uyy = uξξ ξy2 + 2uξη ξy ηy + uηη ηy2 + uξ ξyy + uη ηyy .
Substituting all in (11) we get
∂2u ∂2u ∂2u ∂u ∂u

Ā 2
+ B̄ + C̄ 2
+ D̄ + Ē + F̄ u = Ḡ , (12)
∂x ∂x∂y ∂y ∂x ∂y
where
Ā = Aξx2 + Bξx ξy + C ξy2

B̄ = 2Aξx ηx + B (ξx ηy + ξy ηx ) + 2C ξy ηy
C̄ = Aηx2 + Bηx ηy + C ηy2
D̄ = Aξxx + Bξxy + C ξyy + Dξx + E ξy
Ē = Aηxx + Bηxy + C ηyy + Dηx + E ηy
F̄ = F , Ḡ = G .

Now we can verify that
B̄ 2 − 4ĀC̄ = (ξx ηy − ξy ηx )2 (B 2 − 4AC )
and the transformation discussed above does not alter the type of the
PDE. Finally it can be shown that following some simple steps, one can
determine specific forms of ξ and η, so that the PDE (11) takes the
following three cannonical forms
1 uξξ − uηη = φ(ξ, η, u, uξ , uη ) or uξη = φ(ξ, η, u, uξ , uη ) in the
hyperbolic case.
2 uηη = φ(ξ, η, u, uξ , uη ) or uξξ = φ(ξ, η, u, uξ , uη ) in the parabolic case.
3 uξξ + uηη = φ(ξ, η, u, uξ , uη ) in the elliptic case.
In the following slide we discuss each cases separately.

Lecture 7: 09.10.2020

Cannonical Forms for Hyperbolic Equations
Since the discriminant B̄ 2 − 4ĀC̄ > 0 for hyperbolic PDE, we choose ξ and η in
such a way that
Ā = 0 and C̄ = 0.
Thus we have
Aξx2 + Bξx ξy + C ξy2 = 0 and Aηx2 + Bηx ηy + C ηy2 = 0.
Rewriting the above equations we get
2 2
ξx ξx ηx ηx
A +B + C = 0 and A +B + C = 0 = 0.
ξy ξy ηy ηy
Solving the above equations we get
√ √
ξx −B + B 2 − 4AC ηx −B − B 2 − 4AC
= and = . (13)
ξy 2A ηy 2A
Now along the curves ξ(x, y ) = c1 and η(x, y ) = c2 we know
dy ξx dy ηx
= − and =−
dx ξy dx ηy
respectively.
Now solving the differential equations
√ √
dy B − B 2 − 4AC dy B + B 2 − 4AC
= and = (14)
dx 2A dx 2A
we can determine the surfaces ξ(x, y ) = c1 and η(x, y ) = c2 which will
give the coordinates ξ and η that will ensure Ā = C̄ = 0. Using these
coordinates we will be able to determine the desired normal form of the
hyperbolic PDE. Note that the equations (14) are called characteristic
equations.

Example
Find the cannonical of the PDE
3uxx + 10uxy + 3uyy = 0.
Solution
Comparing the given equation with the general form we get
A = 3, B = 10, C = 3, D = E = F = G = 0.
Discriminant, B 2 − 4AC = 100 − 36 = 64 > 0. Hence the PDE is of

hyperbolic type for all (x, y ). Characteristic equations are given by
√ √
dy B − B 2 − 4AC 1 dy B + B 2 − 4AC
= = and = = 3.
dx 2A 3 dx 2A
Solving the characteristic equations we get
x
y − 3x = c1 andy − = c2 .
3
Pinaki Pal (NIT Durgapur) MA331; Odd Semester,x2020-21 40 / 54
Now the new coefficients of the PDE are given by
Ā = Aξx2 + Bξx ξy + C ξy2 = 0

64
B̄ = 2Aξx ηx + B (ξx ηy + ξy ηx ) + 2C ξy ηy = −
3
C̄ = Aηx2 + Bηx ηy + C ηy2 = 0
D̄ = Aξxx + Bξxy + C ξyy + Dξx + E ξy = 0
Ē = Aηxx + Bηxy + C ηyy + Dηx + E ηy = 0
F̄ = F = 0, Ḡ = G = 0.
Hence the cannonical form of the given PDE is

64
uξη = 0 or uξη = 0.
3
Integrating the cannonical equation partially with respect to ξ we get
uη (ξ, η) = f1 (η),
where f1 is some arbitrary function of η.

Integrating once again with respect to η we get
Z
uξη = f (ξ) + f1 (η)dη = f (ξ) + g (η),
where f and g both are arbitrary functions of ξ and η. Now the solution of
the PDE in terms of the independent variables x and y is given by
x
u(x, y ) = f (y − 3x) + g (y − ).
3

Example
Reduce the PDE
uxx − 2 sin xuxy − cos2 xuyy − cos xuy = 0
to a cannonical form.
Solution
Comparing the given equation with the general form we get
A = 1, B = −2 sin x, C = − cos2 x, D = 0, E = − cos x, F = G = 0.
Discriminant, B 2 − 4AC = 4(sin2 x + cos2 x) = 4 > 0. Hence the PDE is

of hyperbolic type for all (x, y ). Characteristic equations are given by
√ √
dy B − B 2 − 4AC dy B + B 2 − 4AC
= = − sin x−1 and = = 1−sin x.
dx 2A dx 2A
Solving the characteristic equations we get
y + x − cos MA331;
Pinaki Pal (NIT Durgapur) x = cOdd Semester,−
1 andy x − cos x = c2 .
2020-21 43 / 54
Now the new coefficients of the PDE are given by

B̄ = 2Aξx ηx + B (ξx ηy + ξy ηx ) + 2C ξy ηy = −4
C̄ = Aηx2 + Bηx ηy + C ηy2 = 0
D̄ = Aξxx + Bξxy + C ξyy + Dξx + E ξy = 0
F̄ = F = 0, Ḡ = G = 0.
Hence the cannonical form of the given PDE is
−4uξη = 0 or uξη = 0.
Integrating the cannonical equation partially with respect to ξ we get
uη (ξ, η) = f1 (η),
where f1 is some arbitrary function of η.

Integrating once again with respect to η we get
Z
uξη = f (ξ) + f1 (η)dη = f (ξ) + g (η),
where f and g both are arbitrary functions of ξ and η. Now the solution of
the PDE in terms of the independent variables x and y is given by
u(x, y ) = f (y + x − cos x) + g (y − x − cos x).

Lecture 8: 29.10.2020

Cannonical forms for Prabolic Equations
In this case the Discriminant B̄ 2 − 4ĀC̄ = 0, which can be true if B̄ = 0
and Ā = 0 or C̄ = 0. Now Ā = 0 implies

2
ξx ξx
=⇒ A +B +C =0
ξy ξy
Which further implies

√
ξx −B + B 2 − 4AC B
= =− .
ξy 2A 2A
Now to find the curve which satisfy the above equation we write the
following characteristic equation
dy B
= .
dx 2A

Solving the above equation we get ξ(x, y ) = c.
Now we can show that Ā = 0 implies B̄ = 0.
We know that
B̄ = 2Aξx ηx + B (ξx ηy + ξy ηx ) + 2C ξy ηy
√
= 2Aξx ηx + 2 AC (ξx ηy + ξy ηx ) + 2C ξy ηy
√ √ √ √
= 2( Aξx + C ξy )( Aηx + C ηy )
ξx B
√ √ √ √
Here ξy = − 2A = − 2 2AAC = − √CA =⇒ Aξx + C ξy = 0.
Therefore we get B̄ = 0. So for the above choice of ξ(x, y ), we have
Ā = B̄ = 0. Which further implies B̄ 2 − 4ĀC̄ = 0.
Hence, in this case we can choose the function η as per our wish only with
the restriction
∂(ξ, η)
6= 0.
∂(x, y )

Example
Find a cannonical form of the PDE
x 2 uxx − 2xyuxy + y 2 uyy = e x .
Solution
Here A = x 2 , B = −2xy , C = y 2 , D = E = F = G = 0.
Discriminant
B 2 − 4AC = 4x 2 y 2 − 4x 2 y 2 = 0,
so the PDE is of parabolic type for all x and y .
The characteristic equation is
dy B y
= =− .
dx 2A x
Solving we get
xy = c,
and we choose
ξ = xy .
Now the other coordinate can be chosen as per our wish and here we
choose η = y . For such choice of ξ and η we have

B̄ = 2Aξx ηx + B (ξx ηy + ξy ηx ) + 2C ξy ηy = 0
C̄ = Aηx2 + Bηx ηy + C ηy2 = y 2
D̄ = Aξxx + Bξxy + C ξyy + Dξx + E ξy = −2xy
F̄ = F = 0, Ḡ = G = e x .
Hence the transformed PDE is given by
y 2 uηη − 2xyuξ = e x .
In terms of ξ and η it can be written as

ξ 2ξ 1 ξ
η 2 uηη = 2ξuξ + e η =⇒ uηη = 2
uξ + 2 e η ,
η η
which is the required cannonical form.
Canonical form for Elliptic PDE (B 2 − 4AC < 0)
The characteristic equations are taken to be

√ √
dy B − B 2 − 4AC dy B + B 2 − 4AC
= and = ,
dx 2A dx 2A
solving which we get the complex conjugate co-ordinates α and β. Then
to get the cannonical form we need to take the transformation
1 i
ξ = (α + β), η = (β − α).
2 2
Using the above transformation we’ll get the canonical form of the
elliptic equations as
uξξ + uηη = φ(ξ, η, u, uξ , uη ).

Example
Reduce the PDE uxx + x 2 uyy = 0 to a canonical form.
Solution
Compairing the given PDE with the standard form we get We may write
A = 1, B = 0, C = x 2 , D = E = F = G = 0 and hence
B 2 − 4AC = −4x 2 < 0. Thus, the given PDE is elliptic in nature. The
characteristic equations are given by
√ √
dy B − B 2 − 4AC dy B + B 2 − 4AC
= = −ix and = = ix.
dx 2A dx 2A
2
Solving the characteristic equations we get y + i x2 = C1 and
2
y − i x2 = C2 . The complex characteristic curves are given by
x2 x2
α=y +i , β =y −i .
2 2

Now to get a canonical form, we use the transformation given by
1 i
ξ= (α + β), η = (β − α),
2 2
which imply
x2
ξ = y, η = .
2
For the above co-ordinates we get
Ā = Aξx2 + Bξx ξy + C ξy2 = x 2 ,

B̄ = 2Aξx ηx + B (ξx ηy + ξy ηx ) + 2C ξy ηy = 0,
C̄ = Aηx2 + Bηx ηy + C ηy2 = x 2 ,
D̄ = Aξxx + Bξxy + C ξyy + Dξx + E ξy = 1,
Ē = Aηxx + Bηxy + C ηyy + Dηx + E ηy = 0,
F̄ = F = 0, Ḡ = G = 0.
Therefore, the required cannonical form is given by

1
x 2 uξξ + x 2 uηη + uξ = 0 =⇒ uξξ + uηη = − uξ .
2η
Exercise
Reduce the following PDEs to cannonical forms
1 uxx + xuyy = 0, x 6= 0.
2 uxx + 2uxy + sin2 (x)uyy + uy = 0.
3 sin2 (x)uxx + sin 22uxy + cos2 (x)uyy + uy = x.
4 uxx + 2uxy + 4uyy + 2ux + 3uy = 0.
5 x 2 uxx − 2xyuxy + y 2 uyy = e x .
y2 x2
6 y 2 uxx − 2xyuxy + x 2 uyy = x ux + y uy .
7 (1 + x 2 )uxx + (1 + y 2 )uyy + xux + yuy = 0.
8 uxx − 2 sin x uxy − cos2 x uyy − cos x uy = 0.
9 uxx + 2xuyy = 0.
10 y 2 uxx − x 2 uyy = 0, x > 0, y > 0.
11 uxx + 2uxy + uyy = 0.
12 e x uxx + e y uyy = u.

Linear PDE With Constant Coefficients: Pinaki Pal

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Linear PDE With Constant Coefficients: Pinaki Pal

Uploaded by

Copyright:

Available Formats

Linear PDE with Constant Coefficients

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 1 / 54

∂nz ∂nz ∂ n z ∂ n−1 z ∂ n−1 z ∂ n−1 z

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 2 / 54

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 3 / 54

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 4 / 54

Liner nonhomogeneous PDE with constant coefficients

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 5 / 54

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 6 / 54

F (D, D 0 )z = f (x, y ), (4)

where f (D, D 0 ) = D n + a1 D n−1 D 0 + · · · + an−1 DD 0n−1 + an D 0n . Now to

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 7 / 54

(mn + a1 mn−1 + a2 mn−2 + · · · + an−1 m + an )φn (y + mx) = 0 (6)

where φ1 , φ2 , . . . are arbitrary functions.

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 8 / 54

x r −1 φ1 (y + mx) + x r −2 φ2 (y + mx) + · · · + φr (y + mx),

while the contributions of the distinct roots to complete solution will

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 9 / 54

where φ1 , φ2 and φ3 are arbitrary functions.

where φ1 , φ2 and φ3 are arbitrary functions.

where φ1 , φ2 , phi3 and φ4 are arbitrary functions.

where φ1 , φ2 , phi3 and φ4 are arbitrary functions.

again and define the Particular Integral (PI) of as follows.

The function u is such that the action of the operator F (D, D 0 ) on u

Clearly u(x, y ) is a particular function which satisfy the PDE homogeneous

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 12 / 54

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 13 / 54

Now the equation (9) is a Lagrange’s equation whose subsidiary equation

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 14 / 54

and after integration we have to replace c by y + mn x.

we need to repeat the above procedure n number of times. Although the

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 15 / 54

I: f (x, y ) = x m y n , m and n are some positive integers

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 16 / 54

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 17 / 54

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 18 / 54

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 19 / 54

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 20 / 54

where n is the order of the PDE and t = ax + by .

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 22 / 54

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 23 / 54

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 24 / 54

with constant coefficients. Then we know how to find the

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 25 / 54

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 26 / 54

m2 − m = 0 and the roots of it are m = 0, 1.

Therefore complementary function is z = φ1 (y ) + φ2 (y + x).

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 28 / 54

Hence the complete solution is given by

z = φ1 (y + 2x) + φ2 (y − 3x) − y cos x + 2 sin x.

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 30 / 54

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 32 / 54

B(x, y )2 − 4A(x, y )C (x, y )

as discriminant of the equation (11).

∂2u ∂2u ∂2u ∂u ∂u

Ā = Aξx2 + Bξx ξy + C ξy2

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 35 / 54

B̄ 2 − 4ĀC̄ = (ξx ηy − ξy ηx )2 (B 2 − 4AC )

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 36 / 54

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 37 / 54

Pinaki Pal (NIT Durgapur) MA331; Odd Semester, 2020-21 39 / 54

3uxx + 10uxy + 3uyy = 0.

Discriminant, B 2 − 4AC = 100 − 36 = 64 > 0. Hence the PDE is of