Lecture 10: Uncertainty (cont.). Stationary states. Particle on a circle.

L10.1 Uncertainty and eigenstates (15:53)

L10.2 Stationary states: key equations (18:43)

L10.3 Expectation values on stationary states (09:00)

L10.4 Comments on the spectrum and continuity conditions (13:09)

L10.5 Solving particle on a circle (11:05)

L10.1 Uncertainty and eigenstates (15:53)

MITOCW | watch?v=1D4VPbhDy_A
PROFESSOR: This definition in which the uncertainty of the permission operator Q in the state psi. It’s always
important to have a state associated with measuring the uncertainty. Because the uncertainty
will be different in different states. So the state should always be there. Sometimes we write it,
sometimes we get a little tired of writing it and we don’t write it. But it’s always implicit.
So here it is. From the analogous discussion of random variables, we were led to this
definition, in which we would have the expectation value of the square of the operator minus
the square of the expectation value. This was always-- well, this is always a positive quantity.
Because, as claim 1 goes, it can be rewritten as the expectation value of the square of the
difference between the operator and its expectation value.
This may seem a little strange. You’re subtracting from an operator a number, but we know
that numbers can be thought as operators as well. Operator of minus a number acting on a
state is well defined. The operator acts on the state, the number multiplies a state. So this is
well defined. And claim 1 is proven by direct computation.
You certainly indeed prove. You can expand what is inside the expectation value, so it’s Q hat
squared. And then the double product of this Q hat and this number. Now, the number and Q
hat commute, so it is really the double product. If you have A plus B times A plus B, you have
AB plus BA, but if they commute it’s 2AB, so this is minus 2 Q hat Q. Like that.
And then, the last term is the number squared, so it’s plus Q squared. And sometimes I don’t
put the hats as well. And all this is the expectation value of the sum of all these things. The
expectation value of a sum of things is the expectation value of the first plus the expectation
value of the second, plus the expectation value of the next. So we can go ahead and do this,
and this is therefore expectation value of Q squared minus the expectation value of this whole
thing.
But now the expectation value of a number times an operator, the number can go out. And this
is a number, and this is a number. So it’s minus 2 expectation value of Q, number went out.
And then you’re left with expectation value of another Q. And the expectation value of a
number is just the number, because then you’re left within the world of psi star psi, which is
equal to 1. So here is plus Q hat squared.
And these two terms, the second and the third, are the same really. They are both equal to
expectation value of Q squared. They cancel a little bit, and they give you this. So indeed, this
is equal to expectation value of Q squared minus expectation value of Q squared.
So claim 1 is true. And claim 1 shows in particular that this number, delta Q squared, in the
expectation value of a square of something, is positive. We’ll see more clearly in a second
when we have claim number 2. And claim number 2 is easily proven. That’s another
expression for uncertainty.
For claim number 2, we will start with the expectation value of Q minus Q squared, like this,
which is the integral dx psi star of x and t, Q minus expectation value of Q, Q minus
expectation value of Q, on psi.
The expectation value of this thing squared is psi star, the operator, and this. And now, think of
this as an operator acting on all of that. This is a Hermitian operator. Because Q hat is
Hermitian, and expectation value of Q is real. So actually this real number multiplying
something can be moved from the wave function to the starred wave function without any cost.
So even though you might not think of a real number as a Hermitian operator, it is. And
therefore this whole thing is Hermitian. So it can be written as dx. And now you have this whole
operator, Q minus Q hat, acting on psi of x and t. And conjugate. Remember, the operator, the
Hermitian operator, moves to act on psi, and the whole thing [INAUDIBLE]. And then we have
here the other term left over.
But now, you see that you have whatever that state is and the state complex conjugated. So
that is equal to this integral. This is the integral dx of the norm squared of Q hat minus Q hat
psi of x and t squared, which means that thing, that’s its complex conjugate.
So this completes our verification that these claims are true, and allow us to do the last step on
this analysis, which is to show that if you have an eigenstate of Q, if a state psi is an eigenstate
of Q, there is no uncertainty. This goes along with our measurement postulate that says an
eigenstate of Q, you measure Q and you get the eigenvalue of Q and there’s no uncertainty.
In particular, we’ll do it here I think.
If psi is an eigenstate of Q, so you’ll have Q psi equal lambda psi, where lambda is the
eigenvalue.
Now, this is a nice thing. It’s stating that the state psi is an eigenstate of Q and this is the
eigenvalue, but there is a little bit more than can be said. And it is. It should not surprise you
that the eigenvalue happens to be the expectation value of Q on the state psi. Why? Because
you can take this equation and integrate dx times psi star. If you bring that in into both sides of
the equation then you have Q psi equals integral dx psi star psi, and the lambda goes up.
Since my assumption whenever you do expectation values, your states are normalized, this is
just lambda. And by definition, this is the expectation value of Q. So lambda happens to be
equal to the expectation value of Q, so sometimes we can say that this equation really implies
that Q hat psi is equal to expectation value of Q psi times psi.
It looks a little strange in this form. Very few people write it in this form, but it’s important to
recognize that the eigenvalue is nothing else but the expectation value of the operator of that
state. But if you recognize that, you realize that the state satisfies precisely Q hat minus Q on
psi is equal to 0. Therefore, if Q hat minus Q on psi is equal to 0, delta Q is equal to 0. By
claim 2. Q hat minus Q expectation value kills the state, and therefore this is 0. OK then.
The other way is also true. If delta Q is equal to 0, by claim 2, this integral is 0. And since it’s
the sum of squares that are always positive, this state must be 0 by claim 2. And you get that
Q minus Q hat psi is equal to 0. And this means that psi is an eigenstate of Q.
So the other way around it also works. So the final conclusion is delta Q is equal to 0 is
completely equivalent of-- I’ll put in the psi. Psi is an eigenstate of Q.
So this is the main conclusion. Also, we learned some computational tricks. Remember you
have to compute an expectation value of a number, uncertainty, you have these various
formulas you can use. You could use the first definition. Sometimes it may be the simplest. In
particular, if the expectation value of Q is simple, it’s the easiest way.
So for example, you can have a Gaussian wave function, and people ask you, what is delta of
x of the Gaussian wave function? Well, on this Gaussian wave function, you could say that
delta x squared is the expectation value of x squared minus the expectation value of x
squared.
What is the expectation value of x? Well, it would seem reasonable that the expectation value
of x is 0. It’s a Gaussian centered at the origin. And it’s true. For a Gaussian it would be 0, the
expectation value of x. So this term is 0. You can also see 0 because of the integral. You’re
integrating x against psi squared. Psi squared is even, x is odd with respect to x going to
minus x. So that integral is going to be 0. So in this case, the uncertainty is just the calculation
of the expectation value of x squared, and that’s easily done. It’s a Gaussian integral.
The other good thing about this is that even though we have not proven the uncertainty
principle in all generality. We’ve only [? multivated ?] it. It’s precise with this definition. So when
you have the delta x, delta p is greater than or equal to h bar over 2, these things are
computed with those definitions. And then it’s precise. It’s a mathematically rigorous result. It’s
not just hand waving. The hand waving is good. But the precise result is more powerful.

L10.2 Stationary states: key equations (18:43)

MITOCW | watch?v=8KQ-yK2xm60
PROFESSOR: We start with the stationary states. In fact, stationary states are going to keep us quite busy for
probably a couple of weeks. Because it’s a place where you get the intuition about solving
Schrodinger’s equation. So the stationary states are simple and useful solutions of the
Schrodinger equation, very nice and simple.
So what are they by definition? Well, we begin with a definition. And the intuition of a stationary
state will follow. See the word stationary is not the same as static. Stationary is something that
maybe it’s kind of moving, but things don’t change. Static is something that’s just not moving.
Stationary states have time dependence. It is very simple, as we will see. So, your definition of
a stationary state has a factorized space and time dependencies.
So this psi of x and t is a stationary state. If you can write it as a product of a function of time
times a function of position. And now, I will try to be careful about this. Wave functions that
have position and time will have this bar at the bottom. Wave functions that don’t have x will
not have it. If I slip on that, please let me know.
So this is a stationary state, but a stationary state has factorized space and time dependencies
and solves the Schrodinger equation-- the solution of Schrodinger’s equation. So what we
need to understand is what this factorized dependence tell us for the Schrodinger equation. So
this stationary state has time dependence.
But the thing that makes them stationary is that if you look at some observable, a Hermitian
operator, and you say, OK, the state has time dependence, so maybe my observable will have
time dependence. No. The observables don’t have time dependence. That is the nice thing
about stationary states.
So, what we call time independent observables which are all observables that are familiar
[INAUDIBLE]-- Hamiltonian, the momentum, the precision, the angular momentum. Time
independent observables have no time dependence. And it kind of looks simple when you
write it like that. Time independent means don’t have time dependence.
But you’ve seen that d dt of the expectation value of x is equal to p over m, or the sum of p
over m, the velocity. And here it is-- a time independent observable that does have time
dependence. So the observable is time independent, but expectation value have no time
dependence in their expectation values.
The time dependence comes from the state-- the state, the psi of x and t have time
dependence, and sometimes it just doesn’t drop out. But for stationary states, it will drop, as
you will see. So, time independent observables have no time dependence in their expectation
values.
OK. So enough of saying things. And let’s just get to them. So we look at the Schrodinger
equation, i h-bar d dt of psi of x and t is equal to h-bar psi of x and t. And just to remind, this
minus h squared over 2m d second d x squared plus V of x. And I will consider states that
have just that at this moment.
Otherwise, it’s not so easy to get time-dependent-- to get stationary states. If you have a
potential that has time dependence, we kind of do the nice thing that we’re going to do. So
we’re going to look only at time independent potentials. So, V of x, like this, times psi of x and
t. OK. So what we do next is to simply substitute the value of the wave function into the
differential equation and see what we get.
So on the left hand side, we will get i h-bar The psi of x goes out but you have d dt, Now a
normal derivative of g of t. And now, this factor, H of psi acts on these two things. Imagine the
function of time times the function of x sitting here.
Well, the function of time can be moved out. So the function of time can be moved across the
Hamiltonian operator. It doesn’t do anything to it. So we’ll have g of t times H-hat of the psi of
x. This is H-hat. And because we had no time dependence in the potential, our assumption,
this whole thing is a function of x.
All right. Next step. Divide this whole equation by the total wave function. Divide by psi. Well, if
you divide by psi, you cancel the middle psi here, and you get the 1 over g. So you get i h-bar
1 over g dg dt is equal-- on the right side, you cancel the g and you get a 1 over psi of x H-hat
psi of x.
And now you look at this equation. And this equation is saying something very strange. The
left hand side is a function of time only. The right hand side is a function of space only. How
can a function of time be equal to a function of space? The only way this can be is if both are
not a function of what they were supposed to be. They’re just numbers. Any function of time
cannot be equal to a function of space, in generality. It just doesn’t make sense.
So each side must be equal to a constant, and it’s the same constant. So each side, this is all
equal to a constant. And we’ll call the constant E. And this E has units of energy. E equal to be
a real constant with units of energy.
You can see the units because the Hamiltonian has units of energy. And whatever psi units it
has-- whatever every unit psi has, they cancel. Here, whatever units g has, they cancel. And hbar over time is units of energy, like in energies equal h-bar omega.
So it has units of energy. The only thing that you may be could say, why real. Quantum
mechanics loves complex numbers. So why don’t we put the complex E? We’ll see what
trouble we get if you choose something that is complex.
So here we go. It’s a real quantity to be-- let’s do it real for the time being. And let’s solve the
first equation. The left hand side, i h-bar dg dt is now equal to gE, or E, where E is a number
and g is a function of time from where g of p is equal to constant E to the minus iEt over h-bar.
Let’s just check it works correctly.
It’s a first order differential equation. Just one function of integration. If you guess the answer,
must be the answer. And that’s the time dependence of a stationary state. It’s exponential
minus iEt over h-bar.
What about the other equation? The other equation has become H psi of x equals E psi of x.
Or, we should write at least once, minus h squared over 2m-- did I make a mistake? No, I
didn’t-- d second dx squared-- I got this normal derivatives here because this is just a function
of x-- plus V of x psi is equal to E psi of x.
This is the same equation that I’m boxing twice, because it’s written in those two ways. And
both ways are very important. And this is part of solving for stationary state. You’ve solved for
g of t. The time dependence was easy to solve for, but the x dependence is complicated, in
general. There, you have to do some work. You have to solve a differential equation. It’s not
that easy.
So many people-- most people-- call this the time independent Schrodinger equation. So that’s
the time independent Schrodinger equation, where H psi equal E psi. And as you can imagine,
solving this differential equation can be challenging, or sometimes very interesting because it
may be that, as far as the first equation is concerned, of what we did here, we don’t know what
this number E is. But it may be that the only reasonable solutions that this equation has are for
some values of E.
The analogy with matrices should tell you that’s probably what’s going to happen. Because
you remember eigenstates and eigenvalues of matrices are peculiar numbers. If you have a
matrix, they’re peculiar eigenvalues. So this equation is an eigenfunction equation. And it’s
possible that it has the solution for some particular values of the energy.
Let me write the whole solution then. If you’ve solved these two things, the whole solution psi
of x and t is now a constant times psi of x times e to the minus iEt over h-bar, where this psi of
x solves this equation. So this is the stationary state.
How about normalizing the stationary state? Can we do that? Well, if we try to normalize it–
psi star of x and t and psi of x and t dx, and you set this equal to 1. This should be the case,
because this should be interesting solutions of the Schrodinger equation. We expect that we
could do particles with them. And we can start wave packets or peculiar states with them.
And let’s see what we get here. Maybe I should know. I’m really [INAUDIBLE]. I’m going to
erase that constant C here. Since we want to normalize this, we will think of this as a
normalization of psi. When we try to normalize psi, we’ll be normalizing middle psi, as you will
see here. There’s no need to put that constant there.
So what do we get here? We get integral dx psi star of x and t, so you have psi of x star. And
now you could say it’s E to the iEt over h-bar. That’s the complex conjugate.
Now, on the other hand, suppose-- I’ll do this this way. [INAUDIBLE] of this other term is psi of
x into the minus iEt over h-bar. And now the good thing about this, you see this integral should
be normalized to 1 to make sense. And it’s a great thing that the time dependence drops out.
And it would not have dropped out if the energy had not been real. If the energy was not real, I
would have had to put here E start. And here I would have had E star minus E and some
function of time. And how can a function of time be equal to 1? Would be a problem. We would
not be able to normalize this wave function.
So E must be real because otherwise we don’t cancel this time dependence, which happily,
when it cancels, it just tells you that the integral dx of psi star of x psi of x must be 1, which is a
very nice thing. So in a stationary state, the normalization condition for a full time dependent
stationary state is that the spacial part is normalized.

L10.3 Expectation values on stationary states (09:00)

MITOCW | watch?v=M2i8R6kMXKA
PROFESSOR: How about the expectation value of the Hamiltonian in a stationary state? You would imagine,
somehow it has to do with energy ion states and energy. So let’s see what happens. The
expectation value of the Hamiltonian on this stationary state. That would be integral dx
stationary state Hamiltonian stationary state. And we’re going to see this statement that we
made a few minutes ago become clear. Well what do we get here? dx psi star of x e to the i Et
over h bar H e to the minus i Et over h bar psi of x. And H hat couldn’t care less about the time
dependence, that exponential is irrelevant to H hat.
That exponential of time can be moved across and cancelled with this one. And therefore you
get that this is equal to dx psi star of x H hat psi of x, which is a nice thing to notice. The
expectation value of H on the full stationary state is equal to the expectation value of H on the
spatial part of the stationary state. That’s neat. I think it should be noted. So it’s equal to the H
of little psi of x. But this one, we can evaluate, because if we are in a stationary state, H hat psi
of x is E times psi of x. So we get an E integral the x psi star of psi, which we already show that
integral is equal to one, so we get the energy.
So two interesting things. The expectation value of this quantity of H in the stationary state is
the same as it’s quotation value of H in the spatial part, and it’s manually equal to the energy.
By the way, you know, these states are energy eigenstates, these psi of x’s, so you would
expect zero uncertainty because they are energy eigenstates. So the zero uncertainty of the
energy operator in an energy eigenstate. There’s zero uncertainty even in the whole stationary
state. If you have an H squared here, it would give you an E squared, and the expectation
value of H is equal to E, so the expectation value of H squared minus the expectation value of
H squared would be zero. Each one would be equal to E squared. Nothing would happen, no
uncertainties whatsoever.
So let me say once more, in general, being so important here is the comment that the
expectation value of any time independent operator, so comments 1, the expectation value of
any time-independent operator Q in a stationary state is time-independent. So how does that
go? It’s the same thing. Q hat on the psi of x and t is general, now it’s integral dx capital Psi of
x and t Q hat psi of x and t equals integral dx-- you have to start breaking the things now. Little
psi star of x E to the i et over H bar. And I’ll put the whole thing here. Q hat Psi of x E to the
minus i et over H.
So it’s the same thing. Q doesn’t care about time So this factor just moves across and cancels
this factor. The time dependence completely disappears. And in this case, we just get-- this is
equal to integral dx psi star Q psi, which is the expectation value of Q on little psi of x, which is
clearly time-independent, because the state has no time anymore and the operator has no
time. So everybody loves their time and we’re in good shape.
The second problem is kind of a peculiarity, but it’s important to emphasize superposition. It’s
always true, but the superposition of two stationary states is or is not a stationary state?
STUDENT: No.
PROFESSOR: No, good. It’s not a stationary state in general because it’s not factorizing. You have two
stationary states with different energies, each one has its own exponential, and therefore, the
whole state is not factorized between space and time. One time-dependence has one spacedependence plus another time-dependence and another space-dependence, you cannot
factor it. So it’s not just a plain fact. So the superposition of two stationary states of different
energy is not stationary.
And it’s more than just saying, OK, it’s not stationary. What it means is that if you take the
expectation value of a time-independent operator, it may have time-dependence, because you
are not anymore guaranteed by the stationary state that the expectation value has no timedependence. That’s how, eventually, these things have time-dependence, because these
things are not [INAUDIBLE] on stationary states. On stationary states, these things would have
no time-dependence.
And that’s important, because it would be very boring, quantum mechanics, if expectation
values of operators were always time-independent. So what’s happening? Whatever you
measure never changes, nothing moves, nothing changes. And the way it’s solved is because
you do have those stationary states that will give you lots of solutions. And then we combine
them. And as we combine them, we can get time-dependence and we can get the most
[INAUDIBLE] equation.

L10.4 Comments on the spectrum and continuity conditions (13:09)

MITOCW | watch?v=gMnQ21-pjOA
PROFESSOR: Would solving this equation for some potential, and since h is Hermitian, we found the results
that we mentioned last time. That is the eigenfunctions of h are going to form an orthonormal
set of functions that span the space. You can expand anything on there. This is what we
proved for a general condition operator to some degree.
So the eigenfunctions form an orthonormal set that spans the space. So you’re going to define
that psi 1 with an E1 and psi 2 with an E2, and then this continues. And this is called the
spectrum of the theory because energy eigenstates are considered the gold standard. If you
want to find solving a theory means finding the energy eigenstates. Because if you find the
energy eigenstates, you can solve, you can write any wave function of superposition of
energetic states and then just let them evolve.
And the energetic states involve easily because they are just stationary states. So the
spectrum of the theory is the collection of numbers that are the allowed energies and of
course, the associated eigenfunctions. So the energies may be many, maybe discrete, maybe
it has a little bit of continuous partners, all kind of varieties. But your task is to find those for
any problem.
So the equation that we’re trying to solve is now re-written. We’re going to try to solve it. So
let’s look at it. It’s a second order differential equation with a potential in general. So we had an
example there. It’s there. It’s boxed. So we’ll write it slightly different, remove the potential to
the right-hand side and get rid of the constants here.
So d x squared is equal to 2m over h squared. So this is the equation we have to solve. So
whenever you have a problem, you may encounter a potential, v of x. And the question is how
bad this potential can be. Well, the potential may be nice and simple, or it may be nice but
then has some jumps. It may have infinite jumps, like a potential is a complete barrier, or it
may have delta functions. all these are v of x equal possibles. All of them.
Many things can happen with a potential. In fact, the potential can be as strange as you’re
one, depending on what problems you want to solve. So it’s your choice.
Now, we’re going to accept, in fact, all of those potentials for our analysis. May be nice and
smooth. There may have discontinuities. It may have infinite discontinuities, and worse things
like delta function. But worse things than that we will ignore, and there are worse things than
that. Maybe a potential discontinues at every point, or maybe a potential has delta functions
and derivatives of delta functions. Or potentials that blow up and do all kinds of things.
And I’m not saying you should never consider that. I’m saying that we don’t know of any very
useful case where you get anything interesting with that. But a conceivable a particular time a
singular potential one day could be used. So we’ll look at these potentials and try to
understand how to set up boundary conditions. And we’re going to worry about basically psi
and how does it behave.
And my first claim is that psi of x has to be continuous. So psi of x cannot jump. The wave
function move along but cannot jump. And the reason is a differential equation. Look, if psi of x
was not continuous, if psi of x was like this, and just had a discontinuity, psi of x equal to x, psi
prime of x would contain a delta function and this is continuity. The derivative is infinite.
And psi double prime of x, the second derivative, would have a derivative of a delta function
which is worse because a delta function, we think of it as a spike that is becoming thinner and
higher, but the derivative of the delta function first goes to infinity and then goes to minus
infinity and then comes back up. It’s much worse in many ways.
And look, if you have this differential equation and psi is not continuous, well, the right-hand
side is not continuous. Or if you have a delta function, then something not continuous, but lefthand side, we’ve had a derivative of a delta function that is nowhere on the right-hand side.
On the right-hand side, the worst that could exist is a delta function in v of x. But the derivative
of a delta function doesn’t exist. So you cannot afford to have a psi that is discontinuous. Psi
has to be continuous.
There’s other ways to argue this. You might put them in your notes, but I’ll leave it like that.
Now how about the next case? I will say the following happens too. Sine prime of x is
continuous unless v of x has a delta function. You see, potentials of delta functions are nice,
they are interesting. We will consider that. Delta functions potentials can be attractive
potentials, repulsive potentials of [INAUDIBLE].
So I claim now that psi prime of x has to also be continuous. Why are we worrying about psi
and psi prime is because you need two conditions whenever you’re going to solve this
differential equation at an interface, you will need to know psi is continuous and psi prime is
continuous because of second-order differential equations.
So suppose psi prime is continuous. Then there is no problem. If psi prime is continuous, the
worse that can happen is that the second derivative is discontinuous. And the second
derivative is discontinuous could happen with a potential of this discontinuous, so one problem
if psi prime is continuous.
But psi prime can fail to be continuous if the potential has a delta function. And let’s see that. If
psi prime is discontinuous, then psi double prime is proportional to a delta function.
If psi prime is discontinuous, double prime is proportional to a delta function. But here psi just
takes some value-- there’s nothing strange about it-- in order to have delta function, which is
psi double prime. To be equal to the right-hand side, v of x must have a delta function. And v
will have a delta function.
So it will be a somewhat similar potential, but we’re going to look at them in about a week from
now. But this will be our guidance to solve problems. The continuity of the wave function and
the continuity of the derivative of the wave function. And for this slightly more complicated
problems in which the potential has a delta function, then you will have a discontinuity in psi
prime, and it will be calculable, and it’s manageable, and it’s all very nice.
Now, we do it a little complicated, and everything is mixed up, but you will see that it’s quite
doable.

L10.5 Solving particle on a circle (11:05)

MITOCW | watch?v=2EV1vJAAo8M
BARTON
ZWIEBACH:
–that has served, also, our first example of solving the Schrodinger equation. Last time, I
showed you a particle in a circle. And we wrote the wave function. And we said, OK, let’s see
what is the momentum of it. But now, let’s solve, completely, this problem.
So we have the particle in the circle. Which means particle moving here. And this is the
coordinate x. And x goes from 0 to L. And we think of this point and that point, identify. We
actually write this as, x is the same as x plus L. This is a strange way of saying things, but it’s
actually very practical. Here is 2L, 3L.
We say that any point is the same as the point at which you add L. So the circle is the whole,
infinite line with this identification, because every point here, for example, is the same as this
point. And this point is the same as that point. So at the end of everything, it’s equivalent to
this piece, where L is equivalent to 0.
It’s almost like if I was walking here in this room, I begin here. I go there. And when I reach
those control panels, somehow, it looks like a door. And I walk in. And there’s another
classroom there with lots of people sitting. And it continues, and goes on forever.
And then I would conclude that I live in a circle, because I have just begun here and returned
to the same point that is there. And it just continues. So here it is. You are all sitting here. But
you are all sitting there. And you are all sitting there, and just live on a circle.
So this implies that in order to solve wave functions in a circle, we’ll have to put that psi of x
plus L is equal to psi of x, which are the same points. And we’ll have 0 potential. V of x equals
0. It will make life simple. So the Hamiltonian is just minus h squared over 2m d second dx
squared.
We want to find the energy eigenstate. So we want to find minus h squared over 2m d second
psi dx squared is equal to E psi. We want to find those solutions.
Now it’s simple, or relatively simple to show that all the energies that you can find are either
zero or positive. It’s impossible to find solutions of this equation with a negative energies.
And we do it as follows. We multiply by dx and psi star and integrate from 0 to L. So we do that
on this equation. And what will we get? Minus h squared over 2m integral psi star of x d dx of d
dx psi of x is equal to E times the integral psi star psi x dx.
And we will assume, of course, that you have things that are well normalized. So if this is well
normalized, this is 1. So this is the energy is equal to this quantity. And look at this quantity.
This is minus h squared over 2m.
I could integrate by parts. If I do this quickly, I would say, just integrate by parts over here. And
if we integrate by parts, d dx of psi of x, we will get a minus sign. We’ll cancel this minus sign,
and will be over.
But let’s do it a little bit more slowly. You can put dx, this is equal to d dx of psi star d psi dx
minus d psi star dx d psi dx. I will do it like this, with a nice big bracket.
Look what I wrote. I rewrote the psi star d second of psi as d dx of this quantity, which gives
me this term when the derivative acts on the second factor. But then I used an extra term,
where the derivative acts on the first factor that is not present in the above line. Therefore, it
must be subtracted out. So this bracket has replaced this thing.
Now d dx of something, if you integrate over x from 0 to L, the derivative of something, this will
be minus h bar squared over 2m psi star d psi dx integrated at L and at 0. And then minus
cancels. So you get plus h squared over 2m integral from 0 to L dx d psi dx squared equal E.
And therefore, this quantity is 0. The point L is the same point as the point 0. This is not the
point at infinity. I cannot say that the wave function goes to 0 at L, or goes to 0, because
you’re going to infinity. No, they have a better argument in this case.
Whatever it is, the wave function, the derivative, everything, is periodic with L. So whatever
values it has at L equal 0 it has-- at x equals 0, it has at x equals L. So this is 0. And this
equation shows that E is the integral of a positive quantity. So it’s showing that E is greater
than 0, as claimed.
So E is greater than 0. So let’s just try a couple of solutions, and solve. We’ll comment on them
more in time. But let’s get the solutions, because, after all, that’s what we’re supposed to do.
The differential equation is d second psi dx squared is equal to minus 2mE over h squared psi.
And here comes the thing. We always like to define quantities, numbers. If this is a number,
and E is positive, this I can call minus k squared psi. Where k is a real number. Because k real,
the square is positive. And we’ve shown that the energy is positive.
And in fact, this is nice notation. Because if you were setting k squared equal to 2mE over h
squared, you’re saying that E is equal to h squared k squared over 2m. So, in fact, the
momentum is equal to hk. Which is very nice notation.
So this number, k, actually has the meaning that we usually associate, that hk is the
momentum. And now you just have to solve this. d second psi dx squared is equal to minus k
squared psi. Well, those are solved by sines or cosines of kx. So you could choose sine of kx,
cosine of kx, e to the ikx. And this is, kind of better, or easier, because you don’t have to deal
with two types of different functions. And when you take k and minus k, you have to use this,
too. So let’s try this. And these are your solutions, indeed. psi is equal to e to the ikx. So we
leave for next time to analyze the [INAUDIBLE] details. What values of k are necessary for
periodicity and how we normalize this wave function.