- GATE: MATHEMATICS - PrevlousYe.

-r

UN IT-IX

Partial Differential Equation

Syllabus:

Method of characteristics for first order linear and quasilinear partial differential equations
Second order partial differential equations in two independent variables: classification an~
canonical forms, method of separation ofvariables for Laplace equation in Cartesian and polar
coordinates, heat and wave equations in one space variable ; Wave equation: Cauchy problem
and d'Alembert formula, domains of dependence and influence, non-homogeneous wave
equation; Heat equation: Cauchy problem; Laplace and Fourier transform methods.

Content

9.0 Sum.m.ary . . . . . . . . . . . . . . . .. . . .. . . .. . . .. . . .. . .. . . . . .. . . .. . . .. . .. .. .. . . .. . . . . . . . .. . . .. . . .. . . .. . . .. . . . .. . . .. . . .. . . .. . . 430-445

9.1 Classification, Formation and Solution of Partial Differential

Equations of Order One .... . ... .... ... ...... .. ........... ... .. .. .. .... ............. . .... ............... . 446-453
9.2 Classification ofintegrals and Charpit's Method ............. ................ ........... 454-454
9.3 Second Order Partial Differential Equation .................. . .............................. . 455-459
9.4 Special Types of Partial Differential Equations -

Heat, Wave and Laplace Equation 460-472

······ ··· ····· · ············ ····························· ·· ·· ····

Q DIPS HOUSE , 28, Jia Sarai, Hauz Khas , Near 1.1.T., New Delhi-110016 ~ 88-00-22-1386
• [email protected] ~ www . dipsacademy. com
1.
PARTIAL DIFFERENTIAL EQUATION

CLASSIFICATION , FORMATION
AND SOLUTION O F PART IA L
DIFFERENTIAL EQUATIO N
· Summa ry

• Quasi Linear Partial Differential Equation

/\ first order partial differential e quation is
sriicl to h" fl quasi-linear partial differential
-
ORDER ON E S OF equation if it is linear in p and q, i .e. if it is of
the form P(x,y,7-)P t Q(x,y,z)q=R(x, y , z).
• Partial Din:erential Equation (PDE):
An . equation
. which cont ains . partial • Non-Linear Partial Differential Equation
Partial differential equations of the form
denvativesofdependentvaiiabl e wit . 1'1 respect
to two or more tha.11. two indeper1.den t vanables
. f (x, y,z,p,q) 0 which do not come under
is called partial differential equa ...
uon. the above three types are said to be non-linear
partial diffrrential equations.
• Order of a Partial Differential E quati on
The order of a partial differenti·al equation.
· ,
1s
defined FORMATION OF PARTIAL DIFFERENTIAL
. . as order of the highest par t 1a · 1
denvative occurring in the partial differential EQUATION
equation. • By Elimination of Arbitrary Constants
Consider an equation
• Degree of a Partial Differential Equation
The degree of a partial differential equation ... ( 1)
F(x,y,z ,a, b) =0
is the degree of the highest order partial
Where a and b denote arbitrary constants.
derivative which occurs in it after the
Let z be regarded as function of two
equation has been rationalized, i.e., made
free from radicals and fractions. ind ependent variables x and y.
Differentiating ( 1) wrt x and y partially, we get
Note: Commonly used notations throughout
the study of partial differential equations are aF aF aF aF
- + p - and - - q - = 0 ... (2)
ax az ay oz
az az az 2
az
s=--
2

p = - , q=- r = - -2 , axay
ax ay ax E liminating two constants a and bfrom three
equations of ( 1) and (2), we shall obtain an
8 2z equation of the forrnf(x,y,z,p,q ) = O which
and t = - -
ay2
is a partial differential equation of the first
order.
CLASSIFICATION OF FIRST ORDER PARTIAL Also, if there are more arbitrary constants than
DIFFERENTIAL EQUATIONS the number of independent variables, the
above procedure of elimination "'-ill give rise
• Linear Partial Differential Equation
t o partial differential equations ofhigher order
A first order partial differential equation is than the first.
said to be linear partial differential equation
if it is linear in p , q and z, i.e., if it is of the • By Elimination of Arbitrary Function ,p :

form p ( x, y) p + Q ( x, y) q = R ( X, y) Z + S ( X, Y) ·
Let ¢(u, v) = O ... ( 1)
where u and v are functions of x, y and z.
• Semi-Linear Partial Differential Equation
A first order partial differential equation is
We_treat z as dependent variable and x and y
as independent variables.
s aid to be a semi-linear partial differential
equatio n if it is linear in p and q but not p q - l
necessarily in z and the coefficient of P and We can find the determinant u
Q are the functio ns of x and y only i.e. ifit is ' Uy Uz =0
V, Vy Vz
of the form P( x, y)p + Q (x, y )q = R(x, y ,z ) • whic h gives partial differntial equation.

Q DIPS HOUSE , 28 , Jia Sarai , ~au z _Kha s, N e ar 1.1.T. ,_ N e w Delhi - 110016

e [email protected] Ill www .d1psacademy.com
m 88-00-22-1386
- Or we can do through the usual woy of the
elimination of arbitrary functions.
Note:
•
GATE: MATHEMATICS - Previous Year Solved Papa

¢(u1,"i•···•v,,)-O
Cauchy Problem for First Order Quaai-Lbi-,
PDE or Integral surfaces Passing Throulh
a By eliminating arhitnuy functions, we a Given Curve :
always produce quasi-linet1r piwtinl l...el Pp ._ Qq R be the given equation and let
differential equations.
its Lagrange's auxiliary equations give the
':'1 If the given equation contain n arbitrnry
following two independent solutions
functions then climirlAtion of these n
functions gives dsc to partial u(x, y,z) - c,
differential equation of order n.
... ( 1)
• Lagrange's Method
The general solution of the quasi linear
and v(x,y,z) C2

Then we wish to obtain the integral surface

partial differential equation of the form (i. e., a solution) of the given partial differential
P(x,y,z)p+Q(x,y,z)q R(>-.,y,z) ... (1) equa lion which contains the given curve
whose equalion in parametric form are given
is & ( u. u) =0 or u = ¢ ( u) by x = x(t), y = y(t) and z = z(t) where tis a
or u = </J ( u) , where ¢ is an arbitrary function parameter.
Then (1) may be expressed as
and u(x.y,z)=c 1 and u(x,y,z)=c2
are two independent solutions of
U [ X ( J
t) , y (t) , Z ( t) = C l

dx = dy = dz ... (2) and V [ X ( t) , y ( t) , Z ( t)] = C 2 ... (2)

p Q R We eliminate single parameter t from
where c 1 and c 2 are arbitrary constants and equation (2) and get a relation involving c;
at least one of u and u must contain z.

• Generalization of Lagrange's Method:

Finally, we replace c 1 and c 2 with the help of
Let .x;,~, ... , xn be the n independent
(1) and obtain the required integral surface
variables and let
¢(u,v)=0 or ¢(x,y,z)=0.
az az az
P1 =-,P2 =--, ... , p "
8.x; a~
=--
axn • Existence and Uniqueness of Integral
Surface Passing through a given curve
where z is the dependent variable.
Given the Partial differential equation
Consider the general linear partial
differential equation with n independent P(x,y, z) p + Q(x,y, z) q = R(x, y,z) ... ( 1)
variables.
... (1)
with the initial c urve r: (x0 (t), Yo (t), z0 (t))
then there exist a unique solution of the given
where Pi, P2 , ... , P,, are function of .x;, x 2 , •.• , x" . Partial differential equation if and only if
Let U, =C1,U2 =C2,···,un =en be any n
P(xo, Yo, Zo) Q (x0 , y 0 , z 0 )
independent integrals of the Lagrange's
_dx_o * d Yo Vt or
auxiliary equations . dt ~

... (2)
P(xo (t),yo (t) ,z0 (t)) Q(x0 (t),y 0 (t),z0 (t))
d = ~
_o dy 0
-o
Then the general solution of (1) is given by
dt dt

Q H auz _Khas, Near 1.1.T., New Delhi-1 10016

DIPS HOUSE, 28, Jia _S arai , _ L! _ _ _
e
[email protected] 1111 www .dipsacademy. com
88 00 22 1386
 'I
Unit-IX: PARTIAL DIFFERENTIAL EQUATION

• Surfaces Orthogonal to gi
of surfaces

I,ct F(x, y,z) = C

a ven system
·Summary

11(/(s),g(s)) G(s)

where G(s) is a c:ontmuously differentiable

-
.•. ( 1)

•.. (1)
represent a fan:uly of su1 faces • wl1C1C C. f11nction Sue h a problem may have no
IS th~
paran1eter. so lution . Hawever, ifa solution exists in some
Suppose we wish to obtain O nr•1ghborhood of the imt1al curve, then such
SY"! 1c'tn of
surfaces which cut each inem bet 0 f ( 1 ) U 1 11g1t
. o sol11tion ran often be determined using the
1
angles. following steps.

Then the direction ratios of the normal at the Stcp-(i): Find functions h(s) and k(s) (if

point ( x, Y, z) to ( 1) which passes through that PO':HjJbh•) SU( h th~t

F(/(s),g( .,) Cl c,),h(s),k(s)) =0

point are of af af
ax' oy' az. G'(s) =h(s)J'(s)+k(s)g'(s) and

Let the surface z = ¢ ( x, y) FP ( f ( s) , g ( s) , G ( s) , h ( s) , k ( s)) g' ( s )

... (2)
cut each surface of {l) at right angles. -Fq (f (s),g(s) ,G( sJ,h(s),k(s))f'(s) at:. 0
Then the normal at (x, y, z) to (2) has ... (2)

Note that if h(s) and k(s) do not exist, then

az '
az ay
direction ratios ax' - l ·
1.e. p,q,-
1
. (1) has no solution. If there are severa choices

for (h (s) ,k(s)) ,then a solution of lll exists

Since normals at (x, y,z) to (1) and (2) are at for each such choice.
right angles, we have S t e p -{ii) : For each fixed s. solve the following
af af af characteristics system for x s .t),y1 s.r).
p-+q---=0
ax ay az z(s,t) ,p(s, t) ,q (s, t) with the given initial

or
of
p-+q-=-
of of ... (3) conditions p(s,O) = h ls). q(s.O) = kts)
ox oy az
where h (s) and k ls) are the functions found
which is of the form Pp + Qq =R • in step (i).
Conversely, we can easily verify that any d
solution of (3) is orthogonal surface of ( 1) • dt x(s, t) = FP ( x (s, t). Y ls, t). z ts. t). p(s. r). q ls, c))

• Cauchy's Method for Solving Non-Linear

d
PDE dt Y (s, t) = F., (xts, t),y (s, r) ,z ls, r),p(s, r). q ts. r))
To solve partial differential equation

F(x,y ,z, z x,zy ) = 0

subject to an appropriate initial condition (i.e.,
z assume prescribed values on some curve). p(s, t), q(s, t)) + q(s, t) F., (xl s, t) .y( s. t),
Let (f (s),g(s)) traces out a regular curve in
z ( s, t)' p ( s, t) 'q ( s, t))
the xy-plane as s varies.
T h is curve is regarded as initial curve. A cl
clt p(s,t) = -[F, (x(s,t),y(s,t),z (s , t ), p(s, t),
solutio n u(x,y) of the following problem
(know n a s C a u c h y's problem). q ( s,t))+ P(s,t)F, ( x ( s,t), y(s, t),z(s, t),

F( x, y , z , z x, z Y) = 0 p (s,t ) , q( s, t))]

---- - ---~-:-=~~~~~ ~:--=----=--=--=-- - -..:. · :

Jia Sarai, Hauz Khas, Near
Q DIPS HOUSE , 28 · • [email protected] ■ www .d1psacademy. com
1.1.T. ,_ New Delhi-110016 111 88 _00 _22 _1386
 GATE: MATHEMATICS - Previous Year Solved Paper

dz-,p(x,y,z)dx+v1(x,y,z)dy ... (4)

d
- q (s. t) = -[FY (x(s, t). y(s, t) ,z ( s, t). p(s, t),
dt Equation (4) is integrable iff [f,g] = 0
a
q(s, t )) + q(s. t) F,. ( x(s, t) ,y(s, t), z ( s, t).
a(f,g) + P a(J,g) + a(f, g) + q a(f,g) - o
i.e.
a(x, p) a(z, p) a( y ,q) a(z, q)
p(s.t),q(s.t))]
A solution of equation (4) is of the form
Step-(iii):Assand tvary,the point (x,y,z ) ,
F(x,y,z,c)=O, where c is an arbitrary
defined by
parameter.
x = x(s, t),
If equations (1) and (2) are compatible then
y = y(s,t), they have a one-parameter family of common
solutions.
and z=z(s,t) ... (3)

traces out the graph of a solution z of (1) in

• Charpit's Method
(General Method of Solving Partial Differential
the x:yz- space, in a neighborhood of the curve Equations of order one but of any degree):
traced out by (f (s),g(s),G(s)). In some Let the given partial differential equation of
first order and non-linear in P and q be
cases, one can use the first two equations in
(3) to solve for sand t in terms of x and y (say f(x, y ,z, p,q) = 0 ... (1)

s = s(x,y) and t = t(x, y)) to obtain a Then, Charpit's auxiliary equations are given

solution z ( x, y) = z ( s ( x, y), t ( x, y)) , for ( x, y) dp dq

by f x + Pfz Jy + qfz
in a neighborhood of the curve (f (s), g(s)) .
...(2)

2. CLASSIFICATION OF INTEGRALS • Working Rule of Charpit's Method

a Step-(i): Transfer all terms of the given
AND CHARPIT'S METHOD equation to LHS and denote the entire
expression by f .
• Compatible System of First-Order Partial
Differential Equations
Step-(ii): Write down the Charpit's
auxiliary equation (2).
Consider the first order partial differential
equations Step-(iii): Using the value of fin step-

f (x, y , Z, p , q) = 0 ... (1)

(i) write down the values of af , aJ , ... ,
and g(x, y, z, p , q) = 0 .. . (2)
ax ay
Equation (1) and (2) are compatible on domain i.e ., f x ,fy, ... etc. occurring in step-(ii)
Dif and put these in Charpit's equation (2).
Sept-(iv): After simplifying the step-(iii),
a(f,g) d .
a p,q ) * 0 on omain D
J = ( ... (3 ) select two proper fractions so that the
resulting integral may come out to be
(Necessary condition) the simplest relation involving at least
one of p and q .
dz = pdx+qdy
Step-(v): The simplest relation of step-
Should be integrable where (sufficient (i v) is solved a long with the given
condition) equation to determine p and q as
p=tf>(x,y,z) and q = I/J'(x,y,z) functions of x, y and z. Put these values
of p and q in dz=pdx+qdy i.e.,

0 DIPS HOUSE, 28, Jta Sarai, Hauz Khas, Near 1.1.T. , N ew Delhi -1 10016 "9 88-00-22-1386
e
[email protected] i;J www .dipsacademy.com
I.Ill
tt-lX: PARTIAL DIFFERENTIAL EQUATION • Summary

dz-¢( x.y,z)d:1. ... v,(

on
:1.
' y ' z) r,y
integration gives 1h.I.'. compl1•t f'
•
" h 1. c h
OllC
.
Wh1l '119 fl port10I 011,rrcn
_,.,..
c
•
ual quouon of order

3
. S11ppos1, th[lt (l) has hccn rJcnved from 1 ),
integral of the given equnlion
by 11s mg som1 m1•thocl; then the integral (l),
The Singular
. .
and General int q1,in 1s mny bt w l11 r 1, Jw s 118 mony arbitrary r:onstants as
obtained m the usual manner. tl1< If 1111 1nrl1 rwnd,•nt v£1rioblcs, 1s rDJled the
Note: Sometimes Charpit's cqun t·ion s givi• r 0111pl< 11 i11t1•v,rol of (3).
rise to p - a and q - b whcr"'- a ,111
' d IJ ore OI<
constants. ln such cases, putting J) = u nnd /\ lwo pnmrnt•t1·r family of solut10ns
... (4)
q == b in the given equation will give the z-F(x,y,a,b)
required complete mtegral. is called a 'complete integral' of (3) if in the
• Solution of the First Order Partial region considered, the rank of the matnx
Differential Equation
For the first order partial differential equation F ya J is two and z satisfies
Fyb
f (x,y,z,p,q) =0 ... ( 1)
equation (3).
A solution of such a equation is given by

z == z(x,y) which is a continuous ly

• General Integral (GI) or General Solution
(GS)
differentiable function of x and y for all
If we take b = ¢(a) in (4) we get a one
(x, y) ED. We can compute p & q from zand
parameter family of solutions of (3) which is
substitute them into equation (1), then the
a sub-family of the two-parameter family (4)
equation reduces to an identity in x and y. A
as
solution z = z(x,y) when interpreted as a
z = f (x,y,a,r/J(a)) ... (5)
surface in three-dimensional space will be
The envelope of (5), ifit exists, is obtained by
called an integral surface of the partial
eliminating 'a' between (5) and
differential equation. There are several types
of solutions (integral surfaces) for the first fa+ fb ¢'(a)= 0 ... (6)
order partial differential equation.
• Complete Integral (CI) or Complete
If (6) can be solved for a, then a = a ( x, y)
Solution (CS) Substituting for ain (5), we obtain an integral
Let us consider a relation surface, as

rp(x,y,z,a,b)=O ... (1) z == f (x,y,a(x,y),¢(a(x,y)))

In which x, y, z are variables such that z is If the function r/J which defines this sub-
dependent on x and y. family is arbitrary, then such a solution is
Differentiating ( 1) partially wrt x and y called a general integral (general solution) f
respectively, we obtain (3 ). 0

Note: Let z = f (x , y , a ) b e a one parameter

family of solution of (3), then env e lope of this
Since there are two arbitrary constants one ~arameter family, if it exists, is also a
(namely a and b) connected by the above three .
solution of (3 ) a nd it c a n b e obtained b
equations, these can be eliminated and a eliminating 'a' between Y
relation is obtained of the form z J(x,y,a) ... (7)
... (3 )
f (x, y ,z, p , q) = 0
and fc, (x,y,a)
-----G~~~~~~~~~~-
Q .. ·':_s)
;;;;;~~~;-::~~--:
DIPS HOUSE. 28, Jia Sarai, Hauz Khas, Near 1.1.T. , New Delhl-1100 16 88
e Ill '9 [email protected] www.dipsacademy.com -00-22-1386
-
Solv ed Paper
GATE: MATHEMATICS - Prev ious Year

;,; ax1 -Q( a)y 1-b

by
Hen ce, the env elop e ·will be give n
d;,; P(a )dx 1-a dy
or
z = G(x ,y) = f (x.y ,a(x ,y))
z P (a) x 1- ay 1- b
⇒
by solv ing
whe re a(x .y) is obta ined from (8)
clf( z ,p,q ) = O
for a in term s of x and y.
Inte gral : tial
Par ticu lar solu tion or Par ticu lar (i.e . Lhe giv en par tial diff eren
• use d in the equ atio n doe s not invo lve x
and y
Wh en a part icul ar func tion ¢ is
par ticu lar expl icitl y)
gen eral solu tion , we obt ain a get
equ atio n, By usin g Cha rpit 's met hod , we
solu tion of the part ial diff eren tial
diff eren t p = aq
diff eren t cho ices of ¢ may give
part icul ar solu tion s of the part ial
diff eren tial The refo re, f (z,a q, q) =0
equ atio n. = aQ( a,z)
tion : or q = Q(a , z) and also p
• Sin gula r Inte gra l or Sing ular Solu
ined by
Ano ther solu tion of (3) can be obta The n, dz= pdx + qdy = Q( a, z) ( adx+
dy)
-par ame ter
find ing the env elop e of the two
The com plet e inte gral is give n by
fam ily z = f (x,y ,a,b ).
and 'b' from f--d _z_ =ax +y+ b
This is obta ined by elim inat ing 'a'
Q(a ,z)
0, fb = 0
the equ atio ns z = f (x,y ,a,b ), fa=
and is call ed the sing ular inte gral
of (3). g(x ,p) = h(y ,q) (sep arab le type )
foun d from
Also , the sing ular inte gral can be By usin g Cha rpit 's met hod , we
get
lf with out
the part ial diff eren tial equ atio n itse
inte gra l i.e. by g(x ,p) =a ⇒ h(y ,q) = a
kno win g any com plet e
atio ns From thes e, we can solv e for p and
q as
elim inat ing p and q from the equ
f (x,y ,z,p ,q) = 0 p=G (a,x ) and q=H (a,y )

fP (x,y ,z,p ,q) = 0 Thu s, dz= pdx + qdy imp lies

fq (x,y ,z,p ,q) = 0 z = f G(a ,x)d x +J H(a ,y)d y +b

eren tial
Som e stan dar d typ es of par tial diff
• equ atio n and the ir solu tion
by usin g (Cla irau t form ) z = px + qy + g(p ,q)
n by
Cha rpit 's met hod : Her e, com plet e inte gra l is give
z =ax + by+ g ( a, b) for, it is a solu tion
~ f(p, q) =0
(i.e. the giv en par tial diff eren
tial
z and the mat rix ( y+g b
x+g a 1 OJ is of rank
equ atio n doe s not invo lve x, Y and 0 1
expl icitl y) two.
By usin g Cha rpit 's met hod we get
p = a or q =a L
3. LIN EA R PA RT IAL DIF FER EN TIA
The n, we solv e N1 '
EQ UA TIO NS WI TH CO NS TA
f(a, q)= O CO EF FIC IEN TS
or f(p, a)= O atio n of

for q=Q (a) or p=P (a)

• Lin ear par tial diff ere ntia l equ
high er ord er with con stan t coef ficie
nts:
of the forfl l
The Part ial Diff eren tial Equ atio ns
Thu s, dz= a dx + Q(a )dy
016 '9 88-0 0-22 -138 6
Khas , Near 1.1.T. , New Delh i-110
0 DIPS HOU SE, 28, Jia Sara i, Hauz com
~ elnfo@ klttur anl.c om ■ www .dlps acad emy.
Unit-IX: PARTIAL DIFFERENTIAL EQUATION -Summary

a" z
[ kc-;-;;+
cX
k,_ C
X
c" z
n la
y
a" z ]
+ ... + k,, - -
oy "
z ¢\(x--1 ~y) 1-¢.z (x + rn..iy)+ ... 1-¢,.(x+m..Y)
-
or z = ¢ii (y 1- r71ix) + ¢.z (y + rn..ix) + ... + IPn (y + m,..x)
L ir-lz L an -lz a" l ] Case 2: If factors are repeated say
+[ 0 ;:, _, n - 1 + l ;:, , n - 2;:, + ••• + L ~
( .'A v X vy n 1 a y" 1
( Dx - 'T½Dy) is n- times repeated then,

[ cz
+ . . . + llfo -ex + M 1 -ay
az] + N Oz =f ( ,x: ,y ) z = r/Ji (y + m,x) + x¢2 (y + m,_x)

+ ... + x"- 1 r/Jn (y + m,_x)

, , , , are th e function of x and
i.- 's L 's M 's
where •--.

y is called general linear partial differential or z = f//1 (y + m,_x) + Yf//2 (y + m,_x)

equation of order 'n'.
+... +y rt - 11//n ( y+m,_x )
If all the coefficients are constant, then it is
called partial differential equation with or z = ,;1 (x + m 1 y) + y,;2 (x + m,_y)
constant coefficients .

• SYD1bolic Form
Partial Differential Equation of the form
+ ... + y"- 1,;" (x + m,_y)

or Z = 771 ( X + m,_y) + XT/2 ( X + m,_y)

[kuD; 1
+ k,_D; · DY + ... + k " D ; ]z = f (x , y)
+ ... + x"-'77n ( X + m,_y)
in which all the t erms are of order 'n', is called
Non-Homogeneous Partial Differential • Particular integral
It is obtained by solving
equation and if f ( x, y ) = 0 , then it is called
Homogeneous Partial Differential equation .

Here , D = - and D
a =-
a
x ax Y 8y
i.e .'
• General Solution
Homogen eous PDE
(G.S.) of Non-

G.S. = z = C .F + P.I. Here, D1 means integration w.r.t xand

where C.F . i s c omplementary function i.e., X

the solution of the homogenous part of the means integration w.r.t y.

partial differential equation and P.I. is the Case-(i): If f(x,y)=tp(ax+f3y) II:
Particular Integral i. e ., the solution of non-
homogenous part of the PDE. Then substitute Dx by a and DY by /3
• Complementary Function
1
If ¢(Dx, D y) can be written as a product of P.I = ¢(a,f3) ff ... f ,; (u). du ... du
n - times
linear factors then complementary function
where u =ax+ /3y, provided ¢(a,f3) *0 '
11
is obtained by solving ¢ ( D x, DY ) z = 0,
If ¢(a,f3) = 0
where ¢(Dx ,DY )z = (Dx - m,Dy )( Dx - m 2Dy)
i.e.'
•--(D -mn D y )z
X

Case-1: If all the factors in ¢(Dx ,DY) are I

I
distinct then

----~~~~~~~ ~~~~-=-=----
0 DIPS HOUSE, 28, Jia Sarai, Hauz _Khas, Near 1.1.T., New Delhi-110016
e
[email protected] • www.dlpsacademy.com
,_
ia
88
-00-22-1386
- - ~
r I /3' g (a:. I'n)
ff . fv, (u) . d11 ... ci11
(h r)r1m, is
•
GATE: MATHEMATICS - Previous Year Solved Paper

General solution of Homogeneous PDE

rr r/> ( D , D,,) can be reduced to linear factors

the solution can be written in the following

y'
or - , Jf ...f v,(11) . du ... d11 ways depending on ¢ (D,.., D,,) ·
rl( a:) g(a,/J) 7n~
Case-(ii): If f(x,y) v1(ax~ fly) e'"' 1'" Case-(i):

Then substitute D, by a and D 11 by fl

P.1 = l e 0
,+1•11, provided ¢(a,/J) # O. -2'.l.x ....!..1..x.
&(a,/1) Then z =e"' ¢i(,81 x -etiY)+e "> ¢2(P2x-~y)

If &(a,,8)=0

i.e., f (x,y) = (/JD, - a Dy)' g(Dx,Dy)

r
then P.I. = X eax+py
r ! /Jr g (a, /3)

r
or y e ax+py
r ! ar g (a, /3)

Case-(iii) :

If f (x,y) = f//(ax + JJy) = sin(ax + JJy)

or cos(ax+ /Jy) Then

Then, replace D; by - a 2
, D ~ by- /32 and -n -~
z = e ~x {<A (/31 x - a 1 y) + xc;1 (,81 x - a 1 y)} + e ,.,·x

denominator is non-zero.
Case-(iv) : If f ( x, y) = xm y"

"1:s. If m =n then expand ¢ ( D x, D Y ), either

. D D
m power of _x or _Y
Dy DX

If m>nthen expand ¢(Dx, Du), in

D
powerof _Y.
D,
• If ¢(Dx,Du)cannot be reduced to linear
If m < n then expand ¢ ( D"', DY ) in power factors : In this case we take a trial solution
z = Ae1'-" + ku , which when substituted in the
given equation gives the relationship
between hand k.

0 DIPS HOUSE, 28, Jia Sarai, Hauz Khas, Near 1.1.T., New Del hi-110016 '9 88-00-22-1386
e
[email protected] ■ www.dipsacademy.com
Unit-IX: PARTIAL DIFFERENTIAL EQUATION -Summary

4• SECOND ORDER PARTIAL

DIFFERENTIAL EQUATION
Characteristic Roots:
The mntrix corresponding to equation (l) is

o,, .. o,n
-
• Q)I
(/ 12

On .. a}n
Let Rr + Ss + Tt. + f ( x, y' z' z ~' , z_.,, ) =0 ... ( 1) A 0 11 a 12 01n

be the second order semi-linear partial

differential equation, then the roots of the
equation
Find all the Eigen values of A by solving
RA 2 + S)..+T=O ... (2 ) IA - Mj = 0' which will have n eigenvalues
are called characteristic roots.
say ..1.,, i = 1,. ,n
• Characteristic Curve:
Characteristic curves of equation ( 1) are Ca■e-(i) : If J., ;, 0 -ti or J"'i < 0 -fi .
given by
Then, equation ( 1) 1s elliptic.
Case-(ii) : If one or more of the A, = 0 •
Then equation ( 1) is parabolic.
where Ai & ~ are characteristic roots of Case-(iii) : If one of }-, > 0 or }.., < 0 and all the
equation (2). remaining have opposite sign.
2
Case-(i): If S - 4RT > 0 then equation (1) is Then equation (1) is hyperbolic.
hyperbolic and equation (2) has two distinct
REDUCTION TO CANONICAL FORM
real roots (say, Ai and ~ (Ai*~)), so we • Steps for Reducing Hyperbolic Equation into
have two real frunily of characteristic curves. its Canonical Form.:
Case-(ii): If S 2 - 4RT = O then equation (1) is Let the given equation
parabolic and equation (2) has one real root Rr + Ss + Tt. + f (x,y,z,p,q) =0
say 2 = (Ai = ~) , so we have one real family
be hyperbolic i.e. s2 - 4RT > 0 .
of characteristic curves. Step-(i): Find two distinct roots of the
Case-(iii): If 5 2 _ 4 RT < o then equation (1) equation R). 2 + S .ii. + T =O say Ai and ~ (real
. 11 · tic and equation (2) has two complex and distinct).
is e ip . . famil of
roots, then we have two unaginary .y Step-(ii): Find the characteristic cuIVes
. . and no real fanuly of
charactenstic curves
u = f,, (x,y) and v = f 2 (x,y)
characteristic curves.
der Partial
• Classification of Secon d O r
. E tion with Three where f,, ( x, y) = c 1 is a solution of dy + Ai = 0
Differential qua dx
Independent Variables
For the partial differential equation of the form
and / 2 (x,y) = c,, is a solution of~+~= 0
n n . a2z ~ b oz _ + CZ
"
L.,._ L.,._ IJ "'X 8X
J ·a X,
" a - - + L . . . ·=l Step-(Ui): Find p, q, r, s, t in terms of u and u
l=l J=l U I J I

... (1) and put in equation

+f (Xi' X2' X3' ... ' Xn ) = 0 =0
Rr + Ss + Tt. + f ( x, y, z, p, q)
b c are the functions
where aiJ = aj; and al), •'
822
a2z 82z W e ge t ouov = f ( u,v,z,zu,z., ) , w h'1ch . the
1s
of Xi,X2,X3, ... ,xn and ax;axj = 8xj8X;
required canonical form .

. H uz Khas, Near 1.1.T., New Delhi-110016 11 88-00-22-1386.

Jia Sarai. a d' d
0 DIPS HOUSE. 28 , • [email protected] ■ www. 1psaca emy.com
 -• Steps for Reducing Parabolic equation into
its Canonical Form
Let the given equation
GATE: MATHEMATICS - Previous Year Solved Pape

r s t in terms of u and u .. _A
Now fiinc I P , q , ' ,
pt,, t h f'SC. vnl11cs in equation
C
"""''"

Rr f Ss+Tt + f( x, y ,z, p,q) = O

Rr +-Ss + Tt + f( '-·,y,z. p,q) O
fPz
t - - + fJ2z
- - ¢ ( u,v,z,z,,,zu ) '
be parabolic i.e. sl 4RT o W e gc i)u2 av)
Step-(i): Find roots of the equation
which is the required canonical form.
R,i~ + s;., + T = O sa~- ti = ..il A
{real and equal). SPECIAL TYPES OF PARTIAL
5.
Step-(ii): Find the charactens lic curves
DIFFERENTIAL EQUATIONS •
u=f,_(x.y) and v=f2 (x,y ) HEAT, WAVE AND LAPLACE
where f,_ (x,y) = c1 is a solution of EQUATION

dy • Heat Equation:
-
dx
+.?. = 0 and f 2 (x,y) = c 2 is choosen such 1
For the heat equation u xx = k u , ; 0 < x < l

c(u,v)
th at --''-----'- :;t: O. with boundruy conditions u ( 0 , t ) = u (l, t ) = O,
o(x,y)
t~O and initial condition u ( x , O) =f(x),
Step-(iii): Find p, q , r , s, t in terms of u and v
and put in equation O sxsl.
The solution is given by
Rr + Ss + Tt + f(x,y,z,p,q) = 0
2
. ( nnx ) --lcn2- ,?r
ao
oz2
u(x, t) = LDn sm - - e 1
We get - - =1//(u, v,z,zu,zu) n =l l
OU 2

or
o2 z
~
ov
= ¢(u,v,z,zu,zu), where Dn =
2
l
f f(x)sin ( nnxJdx
O l
which is the required canonical form.
The solution u(x,t) is unique.
• Steps for Reducing Elliptic Equation into
its Canonical Form • Heat equation when both ends are insulated.
Let given equation For the heat equation

Rr + Ss + Tt + f(x , y,z,p,q) =0

be Elliptic i.e. S 2 - 4RT < o

with boundary conditions
Step-(i): Find characteristic roots of the
equation R.2 2 + S.2 + T =O u x (O,t) = u .., (l,t) = 0
i.e. , .2 =~,Ai (complex roots) and initial condition

Step-(ii): Find characteristic curves u(x,O) = f (x)

J;_ ( x, y) = c1 and f 2 ( x, y) = c2 The solution is given by

dy dy D .., -icr.' ,..21

b y solving dx + ~ = O ; dx + 2i = O u(x,t) =-o + LDn cos(nnx)e--i,-
2 n~l l

Step-(iii): Let u = J;_ + f 2 = J;_ - 2'

2 an
d v
2i
f2
w h ere D 0 =-
l
ff (x) d x
o

0 DIPS HOUSE , 28 , Jia Sarai , Hauz Khas , Near 1.1.T., New D elhi-110016 '9 88-00-22-1386
e
[email protected] II www . dipsacademy. com
Unit
-IX: PARTIAL DIFFERENTIAL EQUATION S
- ummary

The function K( x, y, t)
-
2f,., ""
1 21

Heat Equation with Non-Homogeneoua sin(~)sin(n~y)

• Boundary Conditions T ,, 1
,"

For the heat equation is cf.died the heat kernel and for every
fixed y the kernel K is a solution of the
1
u.._"' =k ½ ; 0 , x < l ; 0 hcAt equation 9nd it satisfies the
Dirichlet r onditions for t,,,. 0 • The
with boundary conditions
I

u(O,t)=c 1 and u(Z,t)=c 2 so lution is ,,(x,t) = f K(x,y,t)J(y)dy

0

and initial condition

u(x,O) = f(x)
/\lso, K ( x,y,t) =K (y,x,t)

i.<'., K is symmetnc
The solution is given by
For the non - homogeneous heat problem
u ( x, t) = V ( x, t) +Ax+ B u , - ku = F ( x, t) 0 < x < l, t > 0
where v ( x, t) is the solution of corresponding with D1nchlet conditions
homogeneous heat equation u(O,t)=u(l,t)=O, t~O

u(x,O) = f(x), 0 S: x ~ l
The solution is
and v(x,O) = g(x) I
u(x,t)= JK(x,y,t)f(y)dy
given by
nnx --
v(x,t) = LDn sin--e
00
-kn2,r2t
12
0

n=I l I I

+J JK(x,y,t -s)F(y,s)dyds
Dn = l2 s' g (x)sm
. - l - dx (nnx) 00

0 ? -kn•-T""',

and A, B are calculated using the given

where K(x,y, t) =-=- > e 1-
l ~
boundary and initial conditions.
• Heat equation in infinite domain
The problem is u 1 = k:uxx,-oo < x < 00
, t > 0 with

u(x, O) = f(x) , -oo < x < oo and its solution is

• Wave Equation of Finite Length
2 a. For the wave equation
U ( X, t) = l "J' f (,; ) exp ( (X - ,; ) ) d,;
2.,r;Jct _,,, 4kt

• Heat Kernel where c E R is called the "vave speed

~ For homogeneous heat problem with boundary conditions

u I - k:u XX = O > O < x < l, t > 0 u ( 0, c) = u {l, t) = o

with the Dirichlet condition and initial conditions

u(O,t)=u(l,t)=O, t;;;::O u(,,O) =f {x); u,lx,0)=0.

u(x,O)=f(x), o~x:'.5:l The solution is given by

Jia Sarai, Hauz Khas, Near I.I. r., New Dalhi-110016 mi

88 00 ~ - - - · - - - - - -- - -
0 DIPS HOUSE, 2 8 , • info@kitluranlcom • www.d1psacaderny com ,_ • ·22- 1386
 TICS _ Previous Year Solved Paper
GATE: MATHEMA

II ---- end initial conditions

,
u(x,t) = ~Dn sin - 1- cos - 1-
(rvrx) (111rct) u( x, 0) ¢(x),u, (x,0) = g(x)
The solution is given by
where, Dn =z2 fl .
J(x).su1 - l -
( 11:'l'X) d,x ct n,rx
. nnct] s. m
o u(x,t) == Z:"' [ D" c on1l's l +E" sin--
n ..,. J
l l
-

For the wave equation

= c 2 u o.:' O< x <l ; t >0
t n1l'C
U ff ~ . n1l"Xft, (s)sin(t - s)~z s
with boundary conditions
+£..Jsm
n=J
1 0
"
u(O,t)=u(l,t)=0 21 ~
and initial conditions
where Dn == zf ¢(x)sin z
0

u(x,0) = 0, ut (x,0) = g(x)

2 I nJt'X
The solution is given by and E ==-f g(x)sin-1-dx
n nJt'C o

"'
u(x,t) = ~En . (n,rx)
sm . (n1rct)
- l - sm -l- 2 I n1l'X
fn ( t) = n1l'C ff (X, t) sin - l - dx
0
1
2 • (n,rx)
where En =--f g(x)sm - l - dx
n1rc 0 • wave Equation for Infinite Length
(D'Alembert Method)
For the wave equation For the following one-dimensional wave
u tt = c 2 u xx , 0 <x <l; t>0 equation utt = c 2 u _= ; - oo < x < oo with initial
with boundary conditions
conditions u(x,0) = f (x) and u1 (x,O) = g(x).
u(0,t)=u(l,t)=0 The solution is given by
and initial conditions
u(x,0) = f (x); u 1 (x,0) = g(x) u ( x, t) = ½[f ( x +ct)+ J (x - ct)J
The solution is given by
l x+ct

"'
u(x,t) = _LSin (n,rx)
-
+-
2c x-ct
g(i;)di; f
n=I [

Note: If u ( x, t) is the solution of the wave

nJt'Ct) E . (n1l'Ct)]
[ D" cos ( -l- + " sm -l- equation utt = c2 u and A B C Dare the
xx ' ' '
vertices of any parallelogram whose sides are
characteristics.
where Dn = 3. J(x)sin(nJt"X)dx
l
J l
O
Then U(A)+U(C) = U(B)+ U(D)

and En . (n,rx)
= -2- J, g ( X ) sm - - dx • Wave equation of a semi-infinite length
nJt"c O l For the wave equation

• Non-Homogeneous Wave Equation Utt = C 2 U xx ; 0 < X < oo, t > 0

For the wave equation with boundary condition
2
utt -c uxx = J(x,t), 0 < x < l ;t > O u ( 0, t) = 0, 0 < t < 00
with boundary conditions and initial conditions
u(O,t)=u(l,t)=0
u(x,O)=J(x), ut(x,0)=0, O<x<oo

Q DIPS HOUSE, 28, Jia _Sara i, _H auz Khas, Near 1.1.T., New Delhi-110016 '9 _ _ _
e [email protected] ■ www.dipsacademy.com 88 00 22 1386
 Unit-IX: PARTIAL DIFFERENTIAL EQUATION

The solution is

. -{½ [
u(x,t) -
u ( x - ct) + u ( x + ct ) ] ,
·Summary

X ~ Ct
f
[J
f( s )ds o In f.;ic,
-
t the vanishing of

1 the integral 1s , ry condition for

. a nccessa
2 [u ( x + ct) - u ( x _ ct)],
X !:. cl
the solution to exist
The Third Boundary Value Problem or
• Non homogeneous wave equation for
infinite length
The Robin Problem.
This involves finding a function u(x,y)
For the wave equation
which is harmonic in D and satisfies
2
½t - c uxx = F ( x, t) ' -00 < X < oo, t > 0
the condition fJu + h ( s) u ( s) = O on B
with the initial conditions an
u(x,O)=f(x), u,(x,O)=g(x), where h( s).:=O and h (s)'t O.
-OO<X<oo
The solution of the cauchy problem is The Fourth Boundary Value Problem:
1 This involves finding a function u ( x, Y)
u(x,t) = 2 [f(x+ct)+ f(x-ct)]
which is harmonic in D and satisfies
the boundary conditions of different
l x+ct l r x+c(r-rl
types on different portions of the
+-
2c
f 9(s)ds+-f
2c
f F( i;,,) d,;d,
x-ct O x-c(r-rl boundary B. For example, u = fi ( s ) on

au
• Laplace's Equation for steady state Heat
flow
on = f 2 (s)
- on B2 , where

Laplace's equation in the two-dimension is

2
given by V u = u = + u YY = 0 and its solution
is called two-dimensional harmonic function .
• Some Important Points to Rem.etnber :

Boundary Value Problems a Suppose that u ( x, y ) is harmonic in a

Let D be the interior of a simple, closed, bounded domain D and continuous in
smooth curve B and f be a continuous i5 =Du B. Thus u attains its
function defined on the boundary B. maximum on the boundary B of D.
a The First Boundary Value Problem or a Suppose that u(x,y) is harmonic in a
The Drichlet Problem
bounded domain D and is continuous on
The problem of finding a harmonic
D = D uB. Then u attains its
function u(x,y) in D such that it minimum on the boundary B of D.
a The solution of the Dirichlet problem, if
coincides with f on the boundary Bis
it exists, is unique.
called the Drichlet problem.
a The solution of the Neumann problem
The Second Boundary Value Problem
is unique up to the addition of a
or The Neumann Problem constant.
This involves finding a function u(x,y) • Dirichlet problem:
In this problem we shall find harmonic
such that l·t i·s harmonic in D and
function U in a domain D such that U is
a .
. -au_ f ( ) on B , (where - n 1s prescribed on the boundary of domain Dwhere
satisfies
an - s 8 D is the region bounded by simple closed
. tive along the curves.
the directional der~va h condition
outward normal) with t e • Dirichlet Problem for a Rectangle:

----~=:::.:--=~---~-::=~i~H~a:u~z~K~ha;s~N~e
0 DIPS HOUSE, 28,
J ' Sara
i/ T~. .~N~ew~D~el~h~i-~1~1~0~0~1;6-°iim9~8~8:-0~0~-;2;2~-1~3~8~6;--------
' a~r~l.~I.~
info@,kltturanl.com Ill www .dipsacademy.com
- The PDE is u°' + u.,., = O , o < x < a , 0 . ._ y .__ b
with the following boundary conditions
GATE: MATHEMATICS - Previous Year Solved Paper

caae-(iv): When the boundary conditions are

u(x,0) 0, u(x,b) = 0; 0:,; x:,; a

I II m lV u(0,y) = 0, u(a,y) = g(y), 0:,; y:,; b

u(x,O) f(x) 0 0 0 The solution is given by
u(x, b) 0 f(x) 0 0
u(O,y) 0 0 g(y) 0
u(a.y) 0 0 0 g(y )

u= + u"Y =0 ; 0 s: x s; a ; O s: y S: b
where D,. = (nt.a) J 2 b
9
•
(y)sm -b- dy
(n,ry )
0
Case-(i): When the boundary conditions are bsinh - b-

u(x,O) = f (x), u(x, b) = 0, 0 s: x:,; a

u(0,y)=0, u(a,y)=0, 0:,;ys:b • Polar Form of Laplace Equation (Dirichlet
Problem for a Circle)
The solution is given by
The Dirichlet problem for a circle of radius a
is to find solution of
u ( x,y ) = ~
L.,,Dn s.mnn
--x s inh -nn ( y- b) ,
n=l a a
v' u
2
=u l'T
+ -1r u r + 2r1 Uee =0, r < a

Where D" = 2 f f(x)sin nnx dx subject to the boundary condition

as1"nh(nnb)
-- 0 a
a u(a,0)=f(0) and it is

Case-(ii): When the boundary conditions are 00

u(r,0) = ___Q_
a + L, -r )" (an cos n0 + b,. sinnB)
(

u(x,0) = 0, u(x,b) = f (x), 0 S: x S: a 2 n =l a

where a 0 ,a,. and b,. are constants given by

u(O,y) = 0, u(a,y) = 0, 0 S: y S: b
The solution is given by l 2nc

00
a,.=-
7r
J f(B)cosn0d0, n = 0, 1, 2, . ..
0
nn x .
u(x,y) = ~Dn sm~smh ( nn )
•
-;;_-Y ,
l 2nc
and b" = - Jf (0)sinn0d0, n = 1, 2, ...

ff
7r 0

Where D" = 2 (x)sin(n1rx)c1x

. h(nnb)
asm -- 0 a • Dirichlet problem for the Upper Half Plane:
a The solution of the problem

Case-(ill): When the boundary conditions are Uxx + Uyy = 0, - 00 < X < oo, y > 0 .

u ( x, 0) = 0, u ( x, b) = 0, 0 s: x s: a u(x,0) = f (x),-oo < x < oo

with the conditions that u is bounded as
u(0,y) = g(y), u(a,y) = 0, 0 s: y S: b
y ➔ oo , u and u x vanish as lxl ➔ oo is
The solution is given by

u(x,y) = f(x)* (2(

'{; y +x
2y 2)

Where D" =
bsinh nna
2
( JJ g(y)sm (nny)
b

o
- - dy
b
•

Q DIPS HOUSE, 28, Jia Sarai, Hauz Khas, Near 1.1.T., New Delhl-110016 1!1 88-00-22-1386
e
[email protected] • www .dlpsacademy.com
 - nr: PARTIAL DIFFERENTIAL EQUATION
U" 11""' -
S
ummary Ill
• l'feumann Problem for the Upper Half Plane:
The solution of the problem
Ux,. + Uyy = 0, - 00 < X < oo, y > Q

u y (x,O)=g(x),-oo<x<oo

with the conditions that u is bounded as

y ➔ co, u and ux vanish as lxl ➔ oo and

j g(x)dx = 0 is

u(x,y) -_2._
2
"'J g (~ )lo g [(~-
( x)2
)2 +y2] d~
:re -«> ~-x +a 2

•••

. . Hauz Khas , Near 1.1.T.,_ New Delhi-110016

0 Dips H OUSE
' 28. eJ1a info@k1
Sara, , tt ani com
ur ·
II www .d1psacademy.com