Lecture 12: Properties of 1D energy eigenstates. Qualitative properties of wavefunctions. Shooting m

L12.1 Nondegeneracy of bound states in 1D. Real solutions (12:36)

L12.2 Potentials that satisfy V(-x) = V(x) (14:18)

L12.3 Qualitative insights: Local de Broglie wavelength (15:52)

L12.4 Correspondence principle: amplitude as a function of position (05:53)

L12.5 Local picture of the wavefunction (12:52)

L12.6 Energy eigenstates on a generic symmetric potential. Shooting method (15:26)

L12.1 Nondegeneracy of bound states in 1D. Real solutions (12:36)

MITOCW | watch?v=EdXaUfRynx8
PROFESSOR: You first are facing the calculation of the energy eigenstate with some arbitrary potential. You
probably want to know some of the key features of the wave functions you’re going to
calculate. So in fact, all of today’s lecture is going to be devoted to this intuitive, qualitative
insights into the nature of the wave function. So we will discuss a few properties that help us
think clearly. And these are two of those properties. I want to begin with them. Then we’ll do a
third one that we have already used, and we will prove it completely. And then turn to the
classical and semi-classical intuition that lets us figure out how the wave function will look.
And that’s a great help for you. Even if you’re solving for your wave function numerically, you
always need to know what the answer should look like. And it’s ideal if before you calculate,
you think about it. And you realize, well, it should have this t properties. And if you find out that
those are not true, well, you will learn something about your intuition and see what was wrong
with it.
So we’re talking about one dimensional potentials, time independent potentials. And a first
statement that is very important, and you will prove in an exercise after spring break, and that
is the fact that one dimensional potentials, when you look at what are called bound states, you
never find degeneracies, energy degeneracies. And this is when x extends from minus infinity
to infinity. You’ve seen already, in the case of a particle in a circle, there are degenerate
energy eigenstates. But if the potential extends to infinity, there is no such things.
Now what is a bound state? A bound state sounds like a complicated concept. But it is not. It
really means an energy eigenstate that can be normalized. Now if an energy eigenstate can
be normalized and you live in the full real line, that the wave function must go to 0 at infinity.
Otherwise you would never be able to normalize it. And if the wave function goes 0 at infinity,
the bound state is some sort of bump in the middle region or something like that. And it
eventually decays. So this is bound by the potential in some way. And that’s basically what we
use to define a bound state. We’ll take it to be that generally. So this is something, this
property, which is very important, is something you will prove.
But now we go to another property. We’ve emphasized forever that the Schrodinger equation
is an equation with complex numbers. And the solutions have complex numbers. And
suddenly, I wrote a few lectures ago a wave function was real. And I was asked, well, how can
it be real? Well, we’ve discussed stationary states in which the full wave function, capital PSI, is
equal to a little psi of x times the exponential of e to the minus i et over h bar. And there in that
exponential, there is complex numbers on this little psi of x in front of that exponential, which is
what we called basically those energy eigenstates. The e to the minus i et over h bar, it’s
understood that little psi of x is the thing we’ve been looking for. And this psi of x solves the
time independent Schrodinger equation h psi equal e psi. And that equation has no complex
number in it.
So little psi of x can be real. And there’s no contradiction. Because the full solution to the time
dependent Schrodinger equation is complex. But here is a statement. With v of x real, the
energy eigenstates can be chosen to be real. And the words can be chosen are very important
here. It means that you may find a solution that is complex, but you need not stick to that
solution. There is always a possibility to work with real solutions.
And what is the way you prove this? This I will put this in the notes. You don’t have to worry
about the proof. You consider the Schrodinger equation for psi. And you show that psi star, the
complex conjugate of psi, solves the same equation that psi solves. And therefore, if psi is a
solution, psi star is a solution with the same energy. That part is very important. Therefore, if
you have two energy eigenstates with the same energy, you can form the sum. That’s still an
energy eigenstate with the same energy. Even formed in difference, that’s still an energy
eigenstate with the same energy. And the sum of psi plus psi star is real. And the difference
psi minus psi star, if you divide by 2i, is real as well. Therefore you can construct two solutions,
the real part of psi and the imaginary part of psi. And both are solutions to the Schrodinger
equation.
So I’ve said in words what is the proof of the first line. It’s that if you have a psi, psi star is also
a solution. Therefore, psi plus psi star and psi minus psi star are solutions. So given a complex
psi, then psi psi of x. Then psi real of x that we define to be psi of x plus psi star of x over 2.
And the imaginary part of the wave function 1 over 2i psi of x minus psi star of x are both
solutions with the same energy as this one has. So these are the two solutions. So far so
good. You don’t like to work with complex psi? No need to work with complex psi. Work with
real psi.
But here comes the second part of the argument, the second sentence. I want you to be alert
that the second sentence is very powerful. It says that if you have a bound state of a one
dimensional potential, more is true. There are no genuinely complex solutions in this case. Any
solution that you will find, it’s not that it’s complex and then you can find the real and imaginary
part. No, any solution that you will find will be basically real. And how can it fail to be real? It
just has a complex number in front of it that you can ignore. So it is a very strong statement.
That the wave function, it’s not that you can choose to work it. You’re forced to do it up to a
phase.
So how is that possible? How is that true? And here is the argument for the second line. If
we’re talking bound states, then these two are real solutions with the same energy. So now
suppose these are bound states. There is a problem if there are two real solutions with the
same energy. They would be degenerate. And property number 1 says there’s no such thing
as degenerate energy bound states. So they cannot be degenerate.
So if you start with a complex psi, and you build these two, they must be the same solution.
Because since there are no degenerate bound states, then psi, I will write it as psi imaginary,
of x must be proportional to psi real of x. And both are real, so the only possibility is that they
are equal up to a constant, where the constant is a real constant.
You see there cannot be degenerate bound states. So the two tentative solutions must be the
same. But that means that the original solution, psi, which is by definition the real part plus i
times the imaginary part, is now equal to psi r plus i times c times psi r again, which is 1 plus ic
times psi r. And that is basically the content of the theorem. Any solution is up to a number,
just the real solution. So you’re not going to find the real solution has non-trivial different real
imaginary parts here. No, just the real solution and a complex number.
Now if you want, you can just write this as e to the i argument of 1 plus ic times square root of
1 plus c squared psi r. And then it’s literally the way it’s said here. The wave function is
proportional to a real wave function up to a phase. So that’s a very neat situation. And
therefore, you should not be worried that we are going to have to assume many times in our
analysis that the bound states were trying to look for are real. And we plot real bound states.
And we don’t have to worry about, what are you plotting? The real part? The imaginary part?
Many times we can just work with real things.

L12.2 Potentials that satisfy V(-x) = V(x) (14:18)

MITOCW | watch?v=QMeKIiufg5s
PROFESSOR: This will be qualitative insights on the wave function. It’s qualitative, and it’s partially
quantitative of course, insights into, let’s say, real energy eigenstates. So whenever you have
a problem and a potential, we have what is called the total energy, the kinetic energy, and the
potential energy. So you have the energy, which is total, equal kinetic energy plus the potential
energy.
Now, the potential energy, as you’ve seen, sometimes depends on position. We did piecewise
continuous end potentials, but they could be more complicated and do funny things. So this is
a function of x. And classically speaking, we speak of the energy. You see in quantum
mechanics, the energy is an observable and is the result of a measurement with a permission
operator. Sometimes there could be uncertainty, sometimes not.
But in classical physics, which this intuition will come from, you have a total energy. It is
conserved. It’s equal to potential energy and kinetic energy. That will also depend on where
the particle is in the potential. Let’s do a very simple case. Coordinate x, a potential. v of x, this
based the potential. v of x, it’s a constant, nothing that complicated.
And suppose you have a total energy. Now, the total energy in classical mechanics is
conserved. So when I draw a line, I’m not implying that is a function of x, that sometimes the
energy is like that. No, it’s just a number there that I fixed. Here is the energy. And then
wherever you move, if the particle, the classical particle, is here, then it has some potential
energy, v of x, and some kinetic energy, k of x, building up the total energy.
Classically, the kinetic energy determines the momentum. The kinetic energy is p squared
over 2m. Now, the kinetic energy is p squared over 2m. In this case, the kinetic energy is a
constant. The momentum will be a constant. And then what we really want to just say
something about is the wave function. Well, but if we note the momentum classically it’s a
momentum p, we can infer the Broglie wavelength of the particle. And that the Broglie
wavelength would be h over p. And that’s for the wave function.
So we should expect a wave function that has a wave length equal to lambda. After all, that is
what the Broglie did. And from the Broglie, you got the Schrodinger equation. The Schrodinger
equation, in fact, says this, if you look at it again. So if you look at the wave function. Well, it
must have wavelength lambda. And therefore, I’m talking about real wave function. So it could
be a cosine or a sine that has that wavelength lambda.
Of course in quantum mechanics, a cosine or a side doesn’t have exactly-- it’s not an
eigenstate of momentum. But it’s an eigenstate state of energy. And we want to plug
eigenstates of energy. So you will have something like that, with that lambda. And that’s
intuition. You go from the diagram to a kinetic energy, from a kinetic energy to a momentum,
from a momentum to a wavelength, and that’s the wavelength of your energy eigenstate. And
maybe it’s a good idea that you try to convince yourself this is true by looking again at the
Schrodinger equation. For this simple case of a constant potential and an energy that is big,
you will find this result very quickly.
But let’s do now a more interesting case, in which here is x. And the potential is a growing
function of x. And there’s a total energy here still. So if you are at some point here, here is the
potential of x. And now, this is k of x. And now comes the interesting thing. You see, as your
particle, or whatever particle, is moving here, the kinetic energy is decreasing as you move to
the right. So the kinetic energy [INAUDIBLE] velocity and slows down, slows down, slows
down.
The kinetic energy is becoming smaller and smaller. Therefore the momentum is becoming
smaller and smaller. And therefore the wavelength, the Broglie wavelength, must be becoming
bigger and bigger. Now that is not exact because you really have to solve the Schrodinger
equation to do this. But intuitively, you know that if a potential is constant, this is absolutely
true.
The kinetic energy, and the momentum, and the Broglie waveform have related in this way. It
will be sort of true, or approximately true, if the potential is not changing that fast. Because
then it’s approximately constant. So there’s a notion the slowly changing potential, in which we
can talk about the k of x that is decreasing as we move to the right, a p of x that is also
decreasing, and a lambda of x that would be increasing, a wave with the Broglie wavelength
that is increasing.
Now I should have written in here, maybe, k of x, p of x, lambda of x. This is decreasing,
decreasing, increasing. So I can plot it here. And I would say, well, I don’t know exactly how
this goes. But maybe it’s the wavelength is small. And then the wavelength is becoming bigger,
something like that. Well, the wavelength’s becoming bigger in the energy eigenstate that you
will find is true. But there’s also the question whether the amplitude of the wave will change or
not. So we’ll answer that in a couple of minutes.
But the Broglie wavelength now is becoming a function of position. Now, you know that solving
the Schrodinger equation now with an arbitrary potential is a difficult thing. With a linear
potentially it’s a difficult problem, in which the exact solution exists in terms of Airy functions
and things like that. So this can only be an approximate statement that the Broglie wavelength
is becoming bigger and bigger, because the momentum is becoming smaller and smaller. But
it’s a very useful statement. And whenever you look at wave functions of potentials, you see
that thing happening. Questions?
Let me draw another diagram that illustrates these issues.
[SIDE CONVERSATION]
So let’s draw a general picture of a potential now, so we can make a few features here. So
here it is. We’ll have a potential that is like this, v of x, maybe some energy, e, and that’s it.
Now what happens classically, well, if your particle has some energy, you know already this
part is v of x. This is k of x. There is a potential energy and kinetic energy. The kinetic energy
cannot become negative classically. So the particle cannot go to the left of this point called xl,
x to the left. So this region, x left, is the classically forbidden.
Similarly on the right, you cannot go beyond here. Because then you would have negative
kinetic energy. So this is an x right. And everything to the right [INAUDIBLE], x right, is also
classical forbidden. These points, x left and x right, are called turning points. Because those
are the points where a particle, a classical particle, if it lives in this potential, has to bounce
back and turn.
As we mentioned, at any general point, you have v of x and k of x. And this point, for example,
is the point with maximum k of x or maximum velocity. This is the point where the particle is
moving the fastest. And it always slows down as it reaches the turning point. Because the
kinetic energy is becoming smaller and smaller. So as we said, if you had a constant potential,
this would be the solution. It’s constant p, constant lambda, nice, simple wave function. If it’s
not constant, well, nothing is guaranteed. But if it’s sufficiently constant or slowly varied, then
you’re in good shape.
Now what is the meaning of slowly varying? The meaning of slowly varying has to be said in a
precise way. And this is what leads eventually to the so-called WKB approximation of quantum
mechanics. Because we’re giving you the first results of this approximation that you can
understand classically how they go. To mean that you have a slowly varying potential, is a
potential whose percentage change is small in the relevant distances. So it’s the change in the
potential over the relevant distance must be small compared to the potential.
But what is a relevant distance? If we use intuition from quantum mechanics, it’s at the Broglie
wavelength at any point. That is what the quantum particle sees. So what we need is that the
change in the potential over at the Broglie wavelength-- take the derivative multiply it by that.
The Broglie wavelength must be much smaller than the potential itself. And notice, of course,
the potential is a function of x. And even lambda is a function of x there at the Broglie
wavelength.
Now, an exact solution will not be a sine or a cosine. So to say has a precise defined
wavelength is an approximation. It is the approximation of slowly varying. But it’s a nice
approximation. And this lambda is the lambda that would come as h over p of x. And h over p
of x is the square root of 2m times the kinetic energy over h squared-- no, it’s just that, the
square root of 2mk. Square root of 2mk of x.
So the idea is that you can roughly say that the Broglie wavelength here is of some value here.
The momentum is small if the Broglie wavelength is large. And so when you draw things, you
adjust that. You say, OK, here, the momentum is large. Therefore the Broglie wavelength is
small. So you write a short wavelength thing. And then it becomes longer wavelength and then
shorter. And you just tried to get some insight into how this thing looks.

L12.3 Qualitative insights: Local de Broglie wavelength (15:52)

MITOCW | watch?v=x_ngaeI00qU
PROFESSOR: Here is x. And here is a. Various copies of the x-axis. For the ground state, what is the lowest
energy state? It’s not zero energy, because n begins with 1. So it’s this. The lowest energy
state is a sine. So the wave function looks like this. This corresponds to sine. 1. Or n equals 1.
The next one corresponds to n equals 2, and begins as a sine. And it just goes up like this.
You add half a wave each time. Remember, we quantize k a with n pi. So each time that you
increase n, you’re adding pi to k a. So the phase, you see, you have sine of kx. So you have
sine that goes from 0 up to k a. And k a is equal to n pi. So you go from 0 to pi, from 0 to 2 pi,
then from 0 to 3 pi. So it would be one up, one down, like that. And I could do one more. This
one would be with four cycles, two, three, four. So this is psi 1, psi 2, psi 3, and psi 4. Wave
functions do more and more things.
So what can we learn from this wave function? There are several things that we need to
understand. So one important thing is that the wave function, these are all normalizable wave
functions. The ground state-- this is the ground state-- has no nodes. A node, in a wave
function, is called the point where the wave function vanishes. But it’s not the endpoints or the
points at infinity, if you could have a range that goes up to infinity. It’s an interior point that
vanishes. And the ground state has no nodes.
So a node, node, so zero of the wave function, not at the end of the domain. And of the
domain. Because if we included that, I would have to say that the ground states has two nodes
already, you’ll see, around 0. But the 0 at the end of the domain should not be counted as a
node. Nodes are the zeros inside. And look. This has no nodes, and it’s a general fact about
states of potentials. The next excited state has one node. It’s here. The next has two nodes,
and then the next is three nodes.
So the number of nodes of the wave function increases in potential. You have more and more
wave functions with higher and higher excited states, and the number of nodes increases one
by one on each solution. That’s actually a theorem that is valid for general potentials that have
bound states. Bound states are states that are normalizable. So the decay at infinity. You see,
a state that is not normalizable, like a plane, where it is not a bound state. It exists all over.
And it’s a general theorem that this phase, this one-dimensional potentials, whenever you
have bound states, the number of nodes increases with the energy of the eigenstate. We will
see a lot of evidence for this as we move along the course, and a little bit of a proof. Not a very
rigorous proof.
The other thing I want to comment on this thing that is extremely important is the issue of
symmetry. This potential for simplicity, to write everything nicely, was written from 0 to a. So all
the wave functions are sine of n pi x over a. But in some ways, it perhaps would have been
better to put the 0 here. And you say, why? What difference does it make? Well, you have a 0
at the middle of the interval, the potential and the domain of the wave function are symmetric
with respect to x going to minus x.
So actually, when you look at the wave functions thinking you can rethink this as an infinite box
from a over 2 to minus a over 2, and the solutions, you just copy them, and you see now, this
line that I drew in the middle, the ground state is symmetric. The next state is anti-symmetric
with respect to the midpoint. The next state is now symmetric. And the following one, antisymmetric.
So this is also a true fact. If you have bound states of a symmetric potential-- I will prove this
one, probably on Wednesday. A symmetric potential is a potential for which V of minus x is V
of x. Bound states of a symmetric potential are either odd or even. This is not a completely
simple thing to prove. We will prove it, but you need, in fact, another result. It will be in the
homework. Not this week’s homework, but next week’s homework. In fact, homework that is
due this week is due on Friday.
So the bound states of a symmetric potential, a potential that satisfies this, are either odd or
even. And that’s exactly what you see here. That’s not a coincidence. It’s a true fact. The
number of nodes increase. And the other fact that is very important, of bound states, of onedimensional potentials-- supremely important fact. No degeneracies. If you have a bound state
of a potential that is either localized like this or goes to infinity, there are no degenerate energy
eigenstates. Each energy eigenstate here, there was no degeneracy.
Now, that is violated by our particle in a circle. The particle in the circle did have degenerate
energy eigenstates. But as you will see, when you have a particle in a circle, you cannot prove
that theorem. This theorem is valid for particles in infinitely-- not in a circle. For x’s that go from
minus infinity to infinity or x’s with vanishing conditions at some hard walls. In those cases, it’s
true.
So, look, you’re seeing at this moment the beginning of very important general results, of very
fundamental general results that allow you to understand the structure of the wave function in
general. We’re illustrating it here, but they are very much, truly now, general potential. So what
are they? For one-dimensional potential unbound states, no degeneracy, number of nodes
increasing one by one. If the potential is symmetric, the wave functions are either even or odd.

L12.4 Correspondence principle: amplitude as a function of position (05:53)

MITOCW | watch?v=79GY-hI_emE
PROFESSOR: There’s one more property of this thing that is important, and it’s something called the
correspondence principle, which is another classical intuition. And it says that the wave
function, and it addresses the question of what happens to the amplitude of the wave function.
It says that the wave function should be larger in the regions where the particle spends more
time. So in this problem, you have the particle going here. It’s bouncing and it’s going slowly
here, it’s going very fast here. So it spends more time here, spends a lot of time here, spends
a lot of time here. So it should be better in these regions and smaller in the regions that
spends little time.
So this was called the correspondence principle, which is a big name for a somewhat vague
idea. But nevertheless, it’s an interesting thing and it’s true as well. So let me explain this a
little more and get the key point about this. So we say, if you have a potential, you have x and
x plus dx, so this is dx, the probability to be found in the x is equal to psi squared dx, and it’s
proportional to the time spent there. So we’ll say that it’s-- we’ll write it in the following way. It’s
proportional to the fraction of time spent in dx. And that, we’ll call little t over the period of the
motion in this oscillation. The classical particle is doing, the period there.
That’s the fraction of time it spends there. Up two factors of 2, maybe, because it spends
going there and there for the whole period, it doesn’t matter, it’s anyway approximate. It’s a
classical intuition expressed as the correspondence principle. So this is equal to dx over v,
over the velocity that positioned the [INAUDIBLE] velocity T. And this is there for dx. And the
velocity is p over m, so the mass over period and the momentum.
So here we go. Here’s the interesting thing. We found that the magnitude of the wave function
should be proportional to 1 over p of x, or lambda over h bar of x. So then the key result is that
the magnitude of the wave function goes like the square root of the position the [INAUDIBLE]
de Broglie wavelength. So if here the de Broglie wavelength is becoming bigger because the
momentum is becoming smaller, the logic here says that yes indeed, in here, the particle is
spending more time here, so actually, I should be drawing it a little bigger.
So when I try to sketch a wave function in a potential, this is my best guess of how it would be.
And you will be doing a lot of numerical experimentation with Mathematica and get that kind of
insight. They position the [INAUDIBLE] de Broglie wavelength as you have, it is a function of
the local kinetic energy. And that’s what it gives for you.
OK so that is one key insight into the plot of the wave function. Without solving anything, you
can estimate how the wave length goes, and probably to what degree the amplitude goes.
What else do you know? There’s the node theorem that we mentioned, again, in the case of
the square well. The ground state, the bounce state, the ground state bounce state is a state
without the node. The first excited state has one node, the next excited state has two nodes,
the next, three nodes, and the number of nodes increase. With that information, it already
becomes kind of plausible that you can sketch a general wave function.

L12.5 Local picture of the wavefunction (12:52)

MITOCW | watch?v=fWCGM2auQPs
PROFESSOR: OK, so, local picture. It’s all about getting insight into how the way function looks. That’s what
we’ll need to get. These comments now will be pretty useful. For this equation you have one
over psi, d second psi, d x squared is minus 2 m over h squared, E minus v of x. Look how I
wrote it, I put the psi back here, and that’s useful.
Now, there’s a whole lot of discussion-- many textbooks-- about how the way function looks,
and they say concave or convex, but it depends. Let’s try to make it very clear how the wave
function looks. For this we need two regions. So, the first case, A, is when the energy minus v
of x is less than 0. The energy is less than v of x, that’s a forbidden region-- as you can see
there-- so it’s a classically forbidden. Not quantum mechanically forbidden, but classically
forbidden.
What is the main thing about this classically forbidden region is that the right hand side of this
equation is positive. Now, this gives you two possibilities. It may be that psi at some point is
positive, in which case the second psi must also be positive, because psi and the second psi
appear here. If both are positive, this is positive. Or, it may be case two, that psi is negative,
and the second psi-- the x squared-- it’s also negative.
Well, how do we plot this? Well, you’re at some point x, and here it is, a positive wave function
seems to be one type of convexity, another type of convexity for a negative, that’s why people
get a little confused about this. There’s a way to see in a way that there’s is no confusion. Look
at this, it’s positive, second derivative positive. When you think of a second derivative positive,
I think personally of a parabola going up. So, that’s how it could look.
The wave function is positive, up, it’s all real. We’re using the thing we proved at the beginning
of this lecture: you can work with real things, all real. So, the wave from here is x, and here
negative. And the negative opening parabola, that’s something they got. So nice. So, the wave
function at any point could look like this if it’s positive, or, it could look like this if it’s negative.
So, it doesn’t look like both, it’s not double value. So, either one or the other. But, this is easy
to say in words, it is a shape that is convex towards the axis. From the axis it’s convex here
and convex there. So, convex towards the axis.
Now, there’s another possibility I want to just make sure you visualize this. Sometimes this
looks funny-- doesn’t mean actually the way function can look like that-- but, it’s funny because
of the following reason. It’s funny because if you imagine it going forever, it doesn’t make
sense because you’re in a classically the forbidden region. And the way function’s becoming
bigger and bigger is going to blow up.
So, eventually something has to happen. But, it can look like this. So, actually what happens is
that when you’re going to minus infinity-- here is x and we use minus infinity-- it can look like
this. This is an example of this piece that is asymptotic, and it’s positive, and the second
derivative is positive. Or, negative and the second derivative is negative. So that’s a left
asymptote.
Or, you could have a right asymptote, and it looks like this. Again, second derivative positive,
positive wave function. Second derivative negative, negative wave function. So, you may find
this at the middle of the potential, but then eventually something has to take over. Or, you may
find this behavior, or this behavior, at plus minus infinity. But, in any case you are in a
classically forbidden, you’re convex towards the axis. That’s the thing you should remember.
On the other hand, we can be on the classically allowed region. So, let’s think of that. Any
questions about the classically forbidden? Classically allowed, B. E minus v of x greater than 0,
classically allowed. On the right hand side of the equation is negative. So, you can have, one,
a psi that is positive, and a second derivative that is negative. Or, two, a psi that is negative,
and a second derivative that this positive.
So, how does that look? Well, positive and second derivative negative, I think of some wave
function as positive, and negative is parabolic like that. And then, negative and second
derivative positive, it’s possible to have this. The wave function there it’s negative, but the
second derivative is positive. These things are not very good-- they’re not very usable
asymptotically, because eventually if you are like this, you will cross these points. And then, if
you’re still in the allowed region you have to shift.
But, this is done nicely in a sense if you put it together you can have this. Suppose all of this is
classically allowed. Then you can have the wave function being positive, the second derivative
being negative, matching nicely with the other half. The second derivative positive, the wave
function negative, and that’s what the psi function is. It just goes one after another. So, that’s
what typically things look in the classically allowed region. So, in this case, we say that it’s
concave towards the axis. That’s probably worth remembering.
So, one more case. The case C, when E is equal to E minus v of x not is equal to 0. So, we
have the negative, the positive, 0. How about when you have the situation where the potential
at some point is equal to the energy? Well, that’s the turning points there-- those were our
turning points. So, this is how x 0 is a turning point.
And, something else happens, see, the right hand side is 0. We have that one over psi, the
second psi, the x squared is equal to 0. And, if psi is different from 0, then you have the
second derivative must be 0 at x not. And, the second derivative being 0 is an inflection point.
So, if you have a wave function that has an inflection point, you have a sign that you’ve
reached a turning point. An inflection point in a wave function could be anything like that.
Second derivative is positive here-- I’m sorry-- is negative here, second derivative is positive,
this is an inflection point. It’s a point where the second derivative vanishes. So, that’s an
inflection point.
And, it should be remarked that from that differential equation, you also get that the second
psi, the x squared, is equal to E minus v times psi, which is constant. And, therefore, when psi
vanishes, you also get inflection points automatically because the second derivative vanishes.
So, inflection points also at the nodes.
Turning point is an inflection point where you have this situation. Look here, you have negative
second derivative, positive second derivative, the point where the wave function vanishes and
joins them is an inflection point as well. Is not the turning point-- turning point are more
interesting-- but inflection points are more generic.

L12.6 Energy eigenstates on a generic symmetric potential. Shooting method (15:26)

MITOCW | watch?v=45M-BtYAcwg
PROFESSOR: Here is your potential. It’s going to be a smooth, nice potential like that. V of x. x, x. And now,
suppose you don’t know anything about the energy eigenstates. Now, this potential will be
assumed to be symmetric. So here is one thing you can do. You can exploit some things that
you know about this potential.
And here’s the wave function that we’re going to try to plot. And we could say the following.
Let’s see. Whatever energy are here, for bound states, I’m going to eventually be in the
forbidden region. So far on the right here, I will be in the forbidden region. And I must meet the
wave function that looks like the forbidden region wave function. And the only possibility is
something like that. You could say it’s the [INAUDIBLE] of 1, but actually, if it’s a [INAUDIBLE]
then I could multiply by minus 1 and use this one conventionally. That wave function is always
going to be like that over there.
On the other hand, very important-- the on the very left, how will the wave function look? Well
it also has to decay, so it can decay like that, or it might be decaying like this. And in fact, you
don’t know until you figure out what’s happening in the middle. It may be decaying like this or
like that. I fix the sign here, so whatever this does it should either end up like that or end up
like that. So these are the guidelines that you have to solve this. Should begin like that, and
we’ll see. And it should be either symmetric or anti-symmetric. This would be the case antisymmetric, this would be the case symmetric.
OK. So let’s draw one, two, three, four lines there. One, two, three, and four. And I don’t know
where the energies lie. I don’t know what is the ground state energy. And I want to give you an
insight into how you can figure out why you get this energy one decision when this happens.
So let’s plot the wave function for the first case. I don’t know if I have a label, but let’s assume
this is E0, E1, E2, E3. Three energies [INAUDIBLE], and here’s the one for e0.
OK. So I begin here, that’s how it goes. And then I go through my Schrodinger equation,
integrate it. You see? Numerical, you can always integrate the Schrodinger equation. And this
should be always in this region, let me-- like this. And it’s growing. And, oops, there should be
turning points here. There should be turning points-- suppose this is-- I’m not going to get this
so well, but it goes like this. And now this should be a turning point.
So I should change to the other type of curvature, curvature down. But what probably will
happen with E0 is that it will switch and it will go like and start to curve, maybe. Well, if it looks
like that, it must match to the development of the odd piece or the even piece. Now, it’s never
going to match with the odd one, so it might be with the even. And yes, it would match turning
point here, but look what has happened here. You got the corner there.
You know, this was turning slowly, and this is starting to turn slowly, but here there is a
discontinuity in the derivative. So this is not the solution. You try, but you fail. But that’s-- right.
Not every energy gives a solution. So they should have matched continuously and derivative
continuously at that point, but it didn’t have enough time to do that.
On the other hand, if we try the next one, maybe. The turning points will be here. Let’s see
what happens. Well, now the forbidden energies are over here, and now you have a turning
point here that-- in here, the curvature is negative, the second derivative’s curvature. And it’s
larger than it was here. Here it was small, here it’s larger. So it’s going to curve faster. Maybe
if you get the E1 right, it will curve enough so this flat here, in which case the other side will
match nicely and you’ve got the solution. So you probably have to go little by little until this
becomes flattened and, boom, you’ve got the solution. Energy eigenstate.
Let’s go a little further. This graph continues there. Now I want to go to E2. How am I going to
do that? I’m going to do it this way. So this was here, this was there. There is the vertical line
here. And for E2, the turning points are even further out. And here is the wave function. And
let’s look at this thing that I have. Now, the turning point in this one corresponds to the E2
turning point. This is E1.
And now this will go in here, we’ll turn, and we will go curve and maybe do something like that.
Because it’s curving more and more, and faster. So by the time you reach here, this is no
good, because this one will be symmetric. You know, you would have an anti-symmetric one
that is no use. But now you don’t have a solution, again. So as you increase the energy, this is
starting to do this, and that is not quite so good. And then when you go to E3, you have a
turning point over here. So maybe in this case it will go up here and it will start turning, and it
will turn enough to just-- this dip go to the origin.
OK. You’re saying no good either, because this is terrible, this continuous. But, ah, you were
supposed to draw the other one as well. The old one is actually perfect for it. So this is dash, it
doesn’t exist, and this one matches here. So by the time the dip-- this is not a solution, but the
dip goes down and down, and eventually goes to zero, it matches with this one. That’s why I
said sometimes you don’t know whether this matches with the one that comes from here or
the one that comes from the bottom.
So there you go. This is an energy eigenstate again. It’s odd and it has one node. And that
gives you the intuition how, as you sort of come from the end and you reach the middle, you
sometimes match things or sometimes don’t match. And explains why you get energy
quantization.
The other way in which you’re going to gain intuition is with the so-called shooting method,
which is the last thing I want to discuss for a minute. So the shooting method in differential
equations is quite nice. Shooting method. Suppose you have a potential that this symmetric
may be something like this-- it doesn’t look very symmetric. It looks a little better now. And you
want to find energy eigenstates.
You do the following. You say, well, the normalization of the energy eigenstates is not so
important. Let’s look for even states. Now, you can look for even or odd states if the potential
is symmetric. Sometimes the potential will have a wall, in which case you have to require a
symmetric potential. It’s easy to solve, as well. But let’s consider the case when the potential is
symmetric and you look for even states.
So what you do is just, say, you pick an energy. Pick some energy E0. And then you put some
boundary condition. You say that the wave function at x equals 0 is 1. And then you say that
the derivative of the wave function at x equals zero is how much? Any suggestion? How much
should it be?
You see, you have a second order differential equation. The second psi is equal to E minus V.
That’s the Schrodinger equation. You need-- the boundary conditions are the value of psi and
the derivative at one point, and then the computer will integrate for you. Mathematica will do it.
But you have to give me the derivative, so what should I put? A number there, 1, 2, 3? Is that
an unknown? What should I pick?
We must put in the 0, because if you had a wave function whose derivative is not 0, and it’s an
even wave function, it would look like this. And there would be a discontinuity in psi prime–
discontinuous. And that’s not possible unless you have a hard wall or you have a delta
function. So you must put this. And then you integrate numerically. Numerically.
And what will happen? Well, if you integrate numerically, the computer is just going to integrate
and see no problem. Basically, it’s just going to do the interview. Ask the computer to calculate
the wave function out to x equal 5, it will calculate it. So the problem is that-- you can see
visually, if you pick some energy, the wave function will do like something, and then will start
blowing up. And then you say, oh, that energy is no good because the wave function won’t be
normalizable.
And then you go back to the computer and change the energy a little bit, and then you will find,
well, maybe this. Now it blows up in the other direction. No good either. But in some energy in
between, as you change, there must be one in which it does the right thing, which is BK.
Somewhere in between. And numerically, you change the value of the energy, you go-- in the
shooting method, when you shoot it either goes up or down.
And you start working within those two numbers to restrict it until you get here. If you have a
solution with five-digit accuracy, it will do this, this, this, and then blow up. If you have a
solution with 10 digits after it, it will do this and go up to here and blow up again. You need
500-digits accuracy to get that wall. But it’s a fun thing that you can do numerically and play
with it. You can calculate five digits accuracy, ten digits accuracy within a matter of minutes.
It’s very practical, and it’s very nice, but one thing you have to do is clean up your equation
before you start. You cannot have an equation in Mathematica with h-bar and m and all that.
So you have to clean up the units, is the first step, and write it as an equation question without
units. Your And this plots very nicely in Mathematica, and you will have lots of practice.