Homework 6

Sixth homework set for PHY-422A/625A: Mathematical Methods II
Instructor: Kaushik Bhattacharya

email: kaushikb@iitk.ac.in
1. Solve the FOLPDE

∂u ∂u
+2 = yex ,
∂x ∂y
where the initial condition is u(x, x) = sin x. Here u is specified on a line x = y. At s = 0
you can take x = y = a and u = sin a.
2. Suppose we have a LSOPDE in n-dimensions, where the variables are xi for i = 1, 2, ···, n.
The LSOPDE is given as:
aij φxi xj + ai φxi + a0 φ = 0 ,
where aij , ai , a0 are all functions of xi . Here we have used the summation convention,
∂φ
where repetitions of indices means a summation over the indices. Here φxi ≡ ∂x i
and
2
φxi xj ≡ ∂x∂i ∂x
φ
j
where both i and j can have values between 1 and n. As the mixed partial
derivatives are equal, i.e., φxi xj = φxj xi we can always assume aij = aji .
(a) We define a coordinate transformation from variables xi to ξk (both i and k run from
1 to n) as
ξ = Bx ,
where ξ is an n component column matrix with components ξk and x is an n com-
ponent column matrix with components xi . Here B is an n × n real matrix. From
this information calculate φxi and φxi xj in terms of ξk variables (using the chain rule
of partial differentiation).
(b) If we define the (symmetric) n × n matrix A ≡ [aij ] and another (symmetric) n × n
matrix Φ ≡ [φxi xj ], then from the last results show that
Tr[AΦ] = Tr[DU ] ,
where the n × n matrix D ≡ BAB T , where B T is the transpose of B.

(c) Justify that D is adiagonalizable if we appropriately choose the matrix B. Can you
say what kind of matrix will diagonalize A?
(d) Suppose you have diagonalized D. Let the eigenvalues of A are called dk . Now show
that if we define new scaled variables as:
ξk
ξ˜k = p ,
|dk |
we can write the original LSOPDE in terms of the variables ξ˜k as:
d˜k φξ̃k ξ̃k + L.O.T = 0 ,
where L.O.T. imples lower order terms and where d˜k is either 0 or ±1.
1
3. Let the LSOPDE in 2-dim, be given as:
aφxx + 2bφxy + cφyy + dφx + eφy + f φ = g ,
where a, b, c, d, e, f, g are all functions of x, y and
∂φ ∂ 2φ
φx ≡ , φxy ≡ ,
∂x ∂x∂y
and so on. The determinant of the above LSOPDE is defined as ∆ = b2 − ac.
(a) First we want to see how the determinant of the system changes under a coordinate
transformation. Suppose we change coordinates from (x, y) to (ξ, η) where ξ =
ξ(x, y) and η = η(x, y). We assume that for this transformation, the determinant of
the Jacobian matrix

ξx ξy
J= ,
ηx ηy
given by |J(x, y)| = 6 0 for all pair of points (x, y). Suppose under these transfor-
mations ω(ξ, η) = φ(x(ξ, η), y(ξ, η)) which also implies φ(x, y) = ω(ξ(x, y), η(x, y)).
From this information we know that: ux = ωξ ξx + ωη ηx and uxx = ωξξ ξx2 + 2ωξη ξx ηx +
ωηη ηx2 + ωξ ξyy + ωη ηxx . Calculate uy , uyy and uxy as done above.
(b) From the above information show that the LSOPDE in terms of ξ and η looks like:
Aωξξ + 2Bωξη + Cωηη + Dωξ + Eωη + F ω = G ,
where
A(ξ, η) = aξx2 + 2bξx ξy + cξy2 ,

B(ξ, η) = aξx ηx + b(ξx ηy + ξy ηx ) + cξy ηy ,
C(ξ, η) = aηx2 + 2bηx ηy + cηy2 .
(c) If we call the matrices obtained from the second order terms in the two cases as:

a b ˜ A B
∆(x, y) = , ∆(ξ, η) = ,
b c B C
then show that

˜ = J ∆ JT ,
∆
where J T is the transpose of the matrix J. From this fact show that both the original
LSOPDE and the transformed LSOPDE have the same discriminant signs, where
the discriminants of the equations are obtained from the second order terms in the
LSOPDE as was shown in class.
(d) Suppose the initial LSOPDE is a Hyperbolic equation and a 6= 0. From this fact can
you justify that you can choose a coordinate transformation such that both A(ξ, η)
and C(ξ, η) are zero. If you look carefully then you will see that A(ξ, η) = 0 and
C(ξ, η) = 0 are essentially the same equations, written in terms of two variables.
2
Henceforth just working with A(ξ, η) = 0 is enough. Show that you can write the
equation A(ξ, η) = 0 as:
(ξx − µ1 ξy )(ξx − µ2 ξy ) = 0 ,
where √
−b ∓ b2 − ac
µ1,2 ≡ .
a
(e) From the last answer, you know that you can obtain two FOLPDEs from the con-
dition A(ξ, η) = 0, and they are:
ξx − µ1 ξy = 0 , ξx − µ2 ξy = 0 .
How will you solve this system using the method of Characteristics? How many
projected characteristic curves will be there? To make the theory consistent you can
associate the first FOPDE (associated with µ1 ) with ξ(x, y) and the second FOPDE
(associated with µ2 ) with η(x, y). Show that both ξ(x, y) and η(x, y) are constants
along the charateristic curves. Here the initial conditions are not given, you can
yourself assume the initial conditions on an initial value curve.
4. Suppose you are given a LSOPDE as:
yφxx + φyy = 0 ,
where we are only interested in the region where y < 0. What kind of a LSOPDE is
this? If we write the above equation in ξ(x, y), η(x, y) coordinates, what will be the
characteristic equations? Solving them can you express ξ and η as functions of x, y? If
the answer is yes, find those functional forms.

Homework 6

Uploaded by

Copyright:

Available Formats

Homework 6

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

Homework 6

Uploaded by

Copyright:

Available Formats

Sixth homework set for PHY-422A/625A: Mathematical Methods II

Instructor: Kaushik Bhattacharya

1. Solve the FOLPDE

where the n × n matrix D ≡ BAB T , where B T is the transpose of B.

d˜k φξ̃k ξ̃k + L.O.T = 0 ,

aφxx + 2bφxy + cφyy + dφx + eφy + f φ = g ,

where a, b, c, d, e, f, g are all functions of x, y and

and so on. The determinant of the above LSOPDE is defined as ∆ = b2 − ac.

Aωξξ + 2Bωξη + Cωηη + Dωξ + Eωη + F ω = G ,

A(ξ, η) = aξx2 + 2bξx ξy + cξy2 ,

then show that

4. Suppose you are given a LSOPDE as:

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.