Lecture 50: Wave equation

IIT Kanpur July 201
20 followers
Sep 26, 2024
Welcome back, we are going to discuss some more application of Fourier transform on arc
in PDE. So, in previous lectures we have seen the application in the heat equation and we
have seen the application in for the Laplacian equation. So, now let us consider the wave
equation, one dimensional wave equation that is u t t is equal to some c times u x x on
r cross time variable t is positive and u of x 0 t equal to 0 this is equal to phi of
x this is on r and u t of x 0 this is equal to psi of x on r. So, now for the time being
let us assume that our phi and psi are nice functions.
So, we know that by D'Alembert's principle formula u of x t is given by 1 by 2 phi of
x minus c t plus phi of x plus c t and plus 1 by 2 c integral x minus c t x plus c t psi
of s ds. So, which means what it says that the solution what at the point x and t what
we need to do suppose this is x then this is x t. So, we are taking the average of this
okay. So, this is what we know. So, now how does one interpret this formula with the help
of Fourier transform that is what we would we are interested to find out.
Now, as we have mentioned we are taking this to be nice class of function. So, this what
we can write is that this is we will write the Fourier inverse and formula for phi this
is equal to minus infinity to infinity phi hat of xi e to the power 2 pi i xi x minus
c t and then d xi plus minus infinity to infinity phi hat of xi e to the power 2 pi
i xi x plus c t d xi and then the other one is 1 by 2 c x minus c t to x plus c t and
then minus infinity to infinity psi s is equal to psi hat at xi e to the power 2 pi i xi
s d xi then ds this is what is the by the Fourier transform.
So, now what one get is that with the simple calculation one sees that this is minus infinity
to infinity phi hat of xi e to the power 2 pi i xi x into the first integral will going
to give you e to the power minus 2 pi i xi c t and the second one is going to give me
2 pi i xi c t then this is d xi this is the first one and then again the second one is
1 by 2 c x minus. So, what we can do is that this is minus infinity to infinity psi hat
at xi and then e to the power this is x minus c to c t to x plus c t e to the power 2 pi
i xi s ds d xi. So, now this is 1 by 2 minus infinity to infinity phi hat of xi e to the
power 2 pi i xi x and then this is 2 times 2 pi xi c t d xi and this is now minus infinity
to infinity. So, if we are writing this then if we integrate this is going to be psi hat
of xi 1 by 2 c psi hat of xi then this is e to the power 2 pi i xi s the s integral
is 2 pi i xi and then this is x minus c t to x plus c t then this is d xi.
So, this becomes 1 by 2 minus infinity to infinity phi hat of xi e to the power 2 pi
i xi x 1 by 2 goes away here. So, we have cos 2 pi xi c t d xi plus now this one is
going to be equal to if I take e to the power xi x plus c t minus e to the power xi x. So,
1 by 2 c minus infinity to infinity psi hat of xi then 1 by 2 pi i xi and then this is
e to the power 2 pi i xi x plus c t minus e to the power 2 pi i xi x minus of c t this
and this phi hat of xi e to the power 2 pi i xi x cos 2 pi xi c t d xi plus this one
is 1 by 4 pi i c minus infinity to infinity psi hat of xi divided by xi and then this
is 2 i sin 2 pi xi c t into e to the power 2 pi i xi x d xi.
So, now i i gets cancelled here. So, this is what we can we can write it down here as
i i goes away. So, this is 2 pi i whatever it is. So, with this calculation what it shows
that we can write down the solution in terms of the cosine and sine. So, actually now here
we need to perform the complex integral. So, with little bit of patience and manipulation
what one can do that suppose every phi hat of xi this is a complex number. So, this is
mod phi hat of xi into e to the power minus 2 pi i some theta xi theta 1 xi this is the
argument and psi hat of xi this is equal to mod of psi hat of xi and e to the power minus
2 pi i theta 1 theta 2 of xi. Now, if I am looking at minus infinity to
infinity phi hat of xi e to the power 2 pi i xi x then this is cos 2 pi xi c t d xi this
I can break this integral minus infinity to 0 plus 0 to infinity and then by the change
of variable here I am going to get 0 to infinity this is going to 0 to infinity phi hat of
xi e to the power 2 pi i xi x cos 2 pi xi c t d xi plus 0 to infinity phi hat of minus
of xi and then e to the power minus 2 pi i xi x cos minus xi is cos xi cos c t because
cos is an even function and what we had noticed is that if we are so calculating phi hat of
minus of xi is equal to phi hat of xi bar. So, therefore, phi hat of minus of xi is the
modulus of this if we plug in this 0 to infinity mod of phi hat of xi and then cos 2 pi xi
c t and then if I add this then I am going to get that this is cos 2 pi xi x minus theta
1 of xi d xi. So, now this is a real valued integral we can calculate similarly we can
calculate the other part of the psi hat of xi.
So, with the similar argument for psi hat we get u of x t this is equal to 1 by 2 or
some constant time mod of phi hat of xi cos 2 pi xi c t then this is cos 2 pi xi x minus
theta 1 xi d xi plus 1 by 2 may be a factor. So, this is 0 to infinity psi mod of psi hat
of xi and then sin 2 pi xi c t and then this is cos 2 pi xi x minus theta 2 xi and divided
by some factor of pi xi c d xi. So, this is the solution of the wave equation u of x t
given by the sin and the cos that means is an infinite super position of right and left
moving planes and the amplitude of each of these waves they are determined by the amplitude
of phi hat of xi and then phi psi hat of xi by xi.
So, the amplitude the wave is determined by amplitude of phi hat of xi and psi hat of
xi by xi. So, this is it and it is a super position of the moving planes. So, that is
what is the interpretation of the solution D'Alembert's formula.
Now, we will look at another application for a different kind of PDE. So, consider the
equation x square u x x plus A x u x which is equal to u t. This is quite similar to
a heat equation except that we do not have this term. So, now this is on r cross 0 infinity
and which is given to us the initial value is f of x on r this is our equation. So, now
what we do is that we will make a change of variable. So, this is a variant this equation
is a variant of heat equation. We make a change of variable substitute x is equal to e to
the power minus of y and now y varies from minus infinity to infinity y is in r.
Therefore, x is always positive and 0 to infinity and therefore, and define u of y or rather
u tilde y t this is equal to u of e to the power minus of y t. So, and denote and f tilde of y is
equal to f of e to the power minus of y this is you denote it. Then if we look at this problem
star reduces to by just taking the derivative will find u double tilde of y plus 1 minus of
a u tilde y this is equal to u tilde of t and u tilde y 0 this is equal to f tilde.
Now, this if a is equal to 1, then this is usual heat equation. Now, we would like to
solve this. So, now if I take the Fourier transform, then what do I get? I get that
by taking the Fourier transform u tilde hat u tilde y y double derivative which is going
to give me minus 4 pi square xi square and then u hat at xi t u tilde hat and from here
this is 1 minus of a into 2 pi i xi and then this is hat of u tilde hat this u tilde hat at xi.
This is equal to u t tilde hat of xi as we have seen that this partial derivative with
respect to 2 and the Fourier transform they commute. So, now what we get is that this
is equal to u tilde t hat at xi. So, this is an ordinary differential equation in t
variable. So, if we solve this, then we are going to get that this will imply this will
imply that u tilde hat at xi t this is equal to some constant times e to the power minus
4 pi square xi square plus 1 minus of a 2 pi i xi whole into t and then the constant
is because we know that what is the value of u tilde hat 0. This is going to be at t
equal to 0, then this is u tilde hat at xi at 0, then this is equal to f tilde hat at
xi. So, this constant is going to be f tilde hat at xi e to the power minus 4 pi square
xi square plus 1 minus of a 2 pi i xi into t. This is what is the solution for this.
So, now if I take this as some function g, then this is going to be f if this is g hat,
then this is f tilde convolution of g of hat, then I can find that u tilde is f tilde convolution
of g. So, now what is that function g? This already we know the Fourier inverse Fourier
transform of this e to the power minus 4 pi square xi square plus 1 minus of a 2 pi i
xi t and then e to the power 2 pi i xi of some this is y d xi because I am trying to
find at the point y and this of course is a known Fourier inversion formula, which is
the some adjustment with the Gaussian and if I take this along with this, so I will
get that u tilde of y, t this is equal to integral of minus infinity to infinity f of
xi f tilde of xi and 1 by square root of 4 pi t by taking the Fourier transform with
dilution of the Gaussian, this is minus of y minus xi plus 1 minus of a t whole square
by 4 t. So, this is d xi, this now take xi is equal to minus of logarithm of n and y
as we have taken minus of logarithm of x, then what do we get u of x t, this is equal
to u tilde of y t, which is equal to minus infinity to infinity, then this is f of eta
and e to the power minus log eta by x because this is xi and then plus 1 minus of a t whole
square divided by 4 t. Derivative is going to give me d eta by eta,
so this is what is the solution of this one variant of the heat equation u xx plus x square
u xx plus ax ux is equal to ut. So, now so there will be other differential equation
what we will be doing it by using Fourier analysis technique, Fourier transform technique
and so one would before coming to those application. So, it will be of interest to know that if
we are what we have notice a beautiful property of the Gaussian that if f of x is equal to
e to the power minus of pi x square, then what we have seen that f hat of xi is equal
to e to the power minus pi xi square. So, now that means if I take from the Schwarz
space to the Schwarz space, then f of f equal to f hat, then this says that actually the
Gaussian this is an Eigen vector for this transform Fourier transform. It will be natural
for us to ask that what are the other Eigen vector of this Fourier transform, that is
one thing what we would like to discuss in the next lecture and another thing what we
would like to be interested to find certain applications of the Fourier transform in statistics
and physics. Thank you.
