MA 201: Lecture - 3
Solving quasilinear PDEs
Quasi-linear First-Order PDEs
A first order quasi-linear PDE is of the form
u u
a(x, y , u) + b(x, y , u) = c(x, y , u). (1)
x y
Quasi-linear First-Order PDEs
A first order quasi-linear PDE is of the form
u u
a(x, y , u) + b(x, y , u) = c(x, y , u). (1)
x y
Let u(x, y ) be a solution of (1).
Quasi-linear First-Order PDEs
A first order quasi-linear PDE is of the form
u u
a(x, y , u) + b(x, y , u) = c(x, y , u). (1)
x y
Let u(x, y ) be a solution of (1).
Setting u = u(x, y ), we obtain a surface S in xyu-plane given by
S := F (x, y , u) = u(x, y ) u = 0.
Quasi-linear First-Order PDEs
A first order quasi-linear PDE is of the form
u u
a(x, y , u) + b(x, y , u) = c(x, y , u). (1)
x y
Let u(x, y ) be a solution of (1).
Setting u = u(x, y ), we obtain a surface S in xyu-plane given by
S := F (x, y , u) = u(x, y ) u = 0.
Note that the gradient vector F = (ux , uy , 1) is normal to the
surface S.
Quasi-linear First-Order PDEs
A first order quasi-linear PDE is of the form
u u
a(x, y , u) + b(x, y , u) = c(x, y , u). (1)
x y
Let u(x, y ) be a solution of (1).
Setting u = u(x, y ), we obtain a surface S in xyu-plane given by
S := F (x, y , u) = u(x, y ) u = 0.
Note that the gradient vector F = (ux , uy , 1) is normal to the
surface S.
Also equation (1) may be written as
aux + buy c = (a, b, c) (ux , uy , 1) = 0. (2)
Quasi-linear First-Order PDEs
A first order quasi-linear PDE is of the form
u u
a(x, y , u) + b(x, y , u) = c(x, y , u). (1)
x y
Let u(x, y ) be a solution of (1).
Setting u = u(x, y ), we obtain a surface S in xyu-plane given by
S := F (x, y , u) = u(x, y ) u = 0.
Note that the gradient vector F = (ux , uy , 1) is normal to the
surface S.
Also equation (1) may be written as
aux + buy c = (a, b, c) (ux , uy , 1) = 0. (2)
This shows that the vector (a, b, c) and the gradient vector F are
orthogonal.
Quasi-linear First-Order PDEs
A first order quasi-linear PDE is of the form
u u
a(x, y , u) + b(x, y , u) = c(x, y , u). (1)
x y
Let u(x, y ) be a solution of (1).
Setting u = u(x, y ), we obtain a surface S in xyu-plane given by
S := F (x, y , u) = u(x, y ) u = 0.
Note that the gradient vector F = (ux , uy , 1) is normal to the
surface S.
Also equation (1) may be written as
aux + buy c = (a, b, c) (ux , uy , 1) = 0. (2)
This shows that the vector (a, b, c) and the gradient vector F are
orthogonal.
In other words, the vector (a, b, c) lies in the tangent plane of the
surface S at each point in the (x, y , u)-space where F 6= 0.
At each point (x, y , u), the vector (a, b, c) determines a direction in
(x, y , u)-space called the characteristic direction.
At each point (x, y , u), the vector (a, b, c) determines a direction in
(x, y , u)-space called the characteristic direction.
Consider a curve in (x, y , u) plane given in parametric form as
x = x(t), y = y (t), and u = u(t), (3)
satisfying
dx
= a(x(t), y (t), u(t)),
dt
dy
= b(x(t), y (t), u(t)), (4)
dt
du
= c(x(t), y (t), u(t)),
dt
At each point (x, y , u), the vector (a, b, c) determines a direction in
(x, y , u)-space called the characteristic direction.
Consider a curve in (x, y , u) plane given in parametric form as
x = x(t), y = y (t), and u = u(t), (3)
satisfying
dx
= a(x(t), y (t), u(t)),
dt
dy
= b(x(t), y (t), u(t)), (4)
dt
du
= c(x(t), y (t), u(t)),
dt
Thus the tangent vector along these curves given by
(dx/dt, dy /dt, du/dt) coincides with the vector field (a, b, c).
At each point (x, y , u), the vector (a, b, c) determines a direction in
(x, y , u)-space called the characteristic direction.
Consider a curve in (x, y , u) plane given in parametric form as
x = x(t), y = y (t), and u = u(t), (3)
satisfying
dx
= a(x(t), y (t), u(t)),
dt
dy
= b(x(t), y (t), u(t)), (4)
dt
du
= c(x(t), y (t), u(t)),
dt
Thus the tangent vector along these curves given by
(dx/dt, dy /dt, du/dt) coincides with the vector field (a, b, c).
The solutions of (4) are called the characteristic curves of the
quasi-linear equation (1).
Remarks.
A smooth union of the characteristic curves given by (4) gives a
surface in (x, y , u)-space having the property that this surface is
tangent to the vector field (a, b, c) at every point. Thus it is an
integral surface.
Remarks.
A smooth union of the characteristic curves given by (4) gives a
surface in (x, y , u)-space having the property that this surface is
tangent to the vector field (a, b, c) at every point. Thus it is an
integral surface.
To solve the IVP (1), pass a characteristic curve through each point
of the initial curve . These curves generate a surface known as
integral surface. This integral surface is the solution of the IVP.
Remarks.
A smooth union of the characteristic curves given by (4) gives a
surface in (x, y , u)-space having the property that this surface is
tangent to the vector field (a, b, c) at every point. Thus it is an
integral surface.
To solve the IVP (1), pass a characteristic curve through each point
of the initial curve . These curves generate a surface known as
integral surface. This integral surface is the solution of the IVP.
The characteristic equations (4) for x and y are not, in general,
uncoupled from the equation for u and hence differ from those in
the linear case.
Remarks.
A smooth union of the characteristic curves given by (4) gives a
surface in (x, y , u)-space having the property that this surface is
tangent to the vector field (a, b, c) at every point. Thus it is an
integral surface.
To solve the IVP (1), pass a characteristic curve through each point
of the initial curve . These curves generate a surface known as
integral surface. This integral surface is the solution of the IVP.
The characteristic equations (4) for x and y are not, in general,
uncoupled from the equation for u and hence differ from those in
the linear case.
The characteristics equations (4) can be expressed in the
nonparametric form as
dx dy du
= = . (5)
a(x, y , u) b(x, y , u) c(x, y , u)
Formation of quasi-linear First-Order PDEs
Theorem
If = (x, y , u) and = (x, y , u) are two given functions of x, y and
u and if G (, ) = 0, where G is an arbitrary function of and , then
u = u(x, y ) satisfies a first order PDE
u (, ) u (, ) (, )
+ = (6)
x (y , u) y (u, x) (x, y )
where
(, ) x y
= .
(x, y ) x y
Remarks.
Since G is an arbitrary function that leads to equation (6), so G is
called the general solution for the equation.
This gives an idea to find the general solution for the quasi-linear
equation
u u
a(x, y , u) + b(x, y , u) = c(x, y , u).
x y
How to prove it?
Differentiate G (, ) = 0 with respect to x and y respectively:
G x + ux u ) + G x + ux u ) = 0, (7)
G y + uy u ) + G y + uy u ) = 0. (8)
Nontrivial solutions for G
= G and G
= G can be found if
x + ux u x + ux u
=0
y + uy u y + uy u
Expanding this determinant gives first-order, quasi-linear equation
(6).
Remarks (contd.)
Consider following quasi-linear equation:
u u
a(x, y , u) + b(x, y , u) = c(x, y , u). (9)
x y
Consider following equation with general solution G (, ) = 0:
u (, ) u (, ) (, )
+ = . (10)
x (y , u) y (u, x) (x, y )
Both equation will be identical if we select = (x, y , u) and
= (x, y , u) in such way that
(, ) (, ) (, )
a= , b= , c= .
(y , u) (u, x) (x, y )
What is (, )?
Theorem (The method of Lagrange)
The general solution of the quasi-linear PDE (1) is
F (, ) = 0, (11)
where F is an arbitrary function, the functions (x, y , u) = c1 and
(x, y , u) = c2 form a solution of the equations
dx dy du
= = . (12)
a b c
Proof. If (x, y , u) = c1 and (x, y , u) = c2 satisfy equation (12) then
the equations
x dx + y dy + u du = 0,
x dx + y dy + u du = 0
are compatible with (12).
Thus, we must have
ax + by + cu = 0,
ax + by + cu = 0.
Solving these equations for a, b and c, we obtain
a b c
(,)
= (,)
= (,)
. (13)
(y ,u) (u,x) (x,y )
Differentiate F (, ) = 0 with respect to x and y , respectively, to have
F u F u
+ + + =0
x u x x u x
F u F u
+ + + = 0.
y u y y u y
F F
Eliminating and from these equations, we obtain
u (, ) u (, ) (, )
+ = . (14)
x (y , u) y (u, x) (x, y )
In view of (13), equation (14) yields
u u
a +b = c.
x y
Thus, we find that F (, ) = 0 is a solution of equation (1). This
completes the proof.
Example
Find the general integral of xux + yuy = u.
Solution. The associated system of equations are
dx dy du
= = .
x y u
From the first two relations, we have
dx dy x
= = ln x = ln y + ln c1 = = c1 .
x y y
Similarly,
du dy u
= = = c2 .
u y y
x
Take (x, y , u) = y and (x, y , u) = yu . The general integral is given by
x u
F ( , ) = 0.
y y
Example
Find the general integral of the equation
u(x + y )ux + u(x y )uy = x 2 + y 2 .
Solution. The characteristic equations are
dx dy du
= = 2 .
u(x + y ) u(x y ) x + y2
Each of these ratios is equivalent to
dx dy du dx dy du
y +x u =0=x y u .
dt dt dt dt dt dt
Consequently, we have
u2 1
d{xy } = 0 and d{ (x 2 y 2 u 2 )} = 0.
2 2
Integrating we obtain two integrals
2xy u 2 = c1 and x 2 y 2 u 2 = c2 ,
where c1 and c2 are arbitrary constants. Take (x, y , u) = 2xy u 2 and
(x, y , u) = x 2 y 2 u 2 .
Hence the general solution is
F (2xy u 2 , x 2 y 2 u 2 ) = 0,
where F is an arbitrary function.
Theorem (Existence and uniqueness result)
Let a(x, y , u), b(x, y , u) and c(x, y , u) in (1) have continuous partial
derivatives with respect to x, y and u variables. Let the initial curve be
described parametrically as
x = x(s), y = y (s), and u(s) = u(x(s), y (s)).
The initial curve has a continuous tangent vector and
dy dx
J(s) = a[x(s), y (s), u(s)] b[x(s), y (s), u(s)] 6= 0 (15)
ds ds
on . Then, there exists a unique solution u = u(x, y ), defined in some
neighborhood of the initial curve , satisfies (1) and the initial condition
u(x(s), y (s)) = u(s).
Example
u u
PDE: u + = 0; IC: u(x, 0) = f (x).
x y
Step 1. (Finding characteristic curves)
To solve the IVP, we parameterize the initial curve as
x = s, y = 0, u = f (s).
The characteristic equations are
dx dy du
= u, = 1, = 0.
dt dt dt
Let the solutions be denoted as x(t, s), y (t, s), and u(t, s). We
immediately find that
u(t, s) = c1 (s), x(t, s) = c1 (s)t + c2 (s), y (t, s) = t + c3 (s),
where ci , i = 1, 2, 3 are constants to be determined using IC.
Step 2. (Applying IC) The initial conditions at t = 0 are given by
x(0, s) = s, y (0, s) = 0, u(0, s) = f (s).
Using these conditions, we obtain
u(t, s) = f (s), x(t, s) = f (s)t + s, y (t, s) = t.
Step 3. (Writing the parametric form of the solution)
The solution is thus given by u(t, s) = f (s) where
x(t, s) = f (s)t + s, y (t, s) = t.
Step 4. (Expressing u(t, s) in terms of u(x, y ))
Applying the condition (15), we find that J(s) = 1 6= 0, along the
entire initial curve. We can immediately solve for s(x, y ) and t(x, y )
to obtain
s(x, y ) = x tf (s) = x yu, t(x, y ) = y .
Thus the solution can also be given in implicit form as
u = f (x yu).
Example
u u
PDE: (y + u) +y = x y ; IC: u(x, 1) = 1 + x.
x y
Observe that the Jacobian
2+s 1
J = = 1 6= 0.
1 0
Therefore, we conclude that there exists an integral surface at least in
the vicinity of .
The characteristics equations and initial conditions are:
dx dy du
(i) = y + u, (ii) = y , (iii) = x y.
dt dt dt
x(0, s) = s, y (0, s) = 1, u(0, s) = 1 + s.
From (ii), we immediately have y (t, s) = e t .
Adding (i) and (iii), we get
d
(u + x) = (u + x) u(t, s) + x(t, s) = (1 + 2s)e t .
dt
From (i), we again obtain
dx
+ x = (2 + 2s)e t x(s, t) = (1 + s)e t e t ,
dt
and
u(t, s) = se t + e t .
Noting that
x y = se t e t ,
we finally get
u(x, y ) = 2/y + (x y ).
Note that the solution is not global (it becomes singular on the x-axis),
but it is well defined near the initial curve.
