Lecture 6: Probability density and current. Hermitian conjugation.

L6.1 Normalizable wavefunctions and the question of time evolution (16:50)

L6.2 Is probability conserved? Hermiticity of the Hamiltonian (20:42)

L6.3 Probability current and current conservation (15:14)

L6.4 Three dimensional current and conservation (18:13)

L6.1 Normalizable wavefunctions and the question of time evolution (16:50)

MITOCW | watch?v=d4skxu7MpFI
BARTON
ZWIEBACH:
We were faced last time with a question of interpretation of the Schrodinger wave function.
And so to recap the main ideas that we were looking at, we derive this Schrodinger equation,
basically derived it from simple ideas-- having operators, energy operator, momentum
operator, and exploring how the de Broglie wavelength associated to a particle would be a
wave that would solve the equation. And the equation was the Schrodinger equation, a free
Schrodinger equation, and then we added the potential to make it interacting and that way, we
motivated the Schrodinger equation and took this form of x and t psi of x and t. And this is a
dynamical equation that governs the wave function.
But the interpretation that we’ve had for the wave function, we discussed what Born said, was
that it’s related to probabilities and psi squared multiplied by a little dx would give you the
probability to find the particle in that little dx at some particular time. So psi of x and t squared
dx would be the probability to find the particle at that interval dx around x.
And if you’re describing the physics of your Schrodinger equation is that of a single particle,
which is the case here-- one coordinate, the coordinate of the particle, this integral, if you
integrate this all over space, must be 1 for the probability to make sense. So the total
probability of finding the particle must be 1, must be somewhere. If it’s in one part of another
part or another part, this probabilities-- for this to be a probability distribution, it has to be wellnormalized, which means 1.
And we said that this equation was interesting but somewhat worrisome, because if the
normalization of the wave function satisfies, if this holds for t equal t-nought then the
Schrodinger equation, if you know the wave function all over space for t equal t-nought, which
is what you would need to know in order to check that this is working, a t equal t-nought, you
take the psi of x and t-nought, integrate it. But if you know psi of x and t-nought for all x, then
the Schrodinger equation tells you what the wave function is at a later time. Because it gives
you the time derivative of the wave function in terms of data about the wave function all over
space.
So automatically, the Schrodinger equation must make it true that this will hold at later times.
You cannot force the wave function to satisfy this at all times. You can force it maybe to satisfy
at one time, but once it satisfies it at this time, then it will evolve, and it better be that at every
time later, it still satisfies this equation.
So this is a very important constraint. So we’ll basically develop this throughout the lecture
today. We’re going to make a big point of this trying to explain why the conditions that we’re
going to impose on the wave function are necessary; what it teaches you about the
Hamiltonian, we’ll teach you that it’s a Hermitian operator; what do you learn about probability-

• you will learn that there is a probability current; and all kinds of things will come out of taking
  seriously the interpretation of this probability, the main point being that we can be sure it
  behaves as a probability at one time, but then for later times, the behaviors and probability the
  Schrodinger equation must help-- must somehow be part of the reason this works out. So
  that’s what we’re going to try to do.
  Now, when we write an equation like this, and more explicitly, this means integral of psi star of
  x and t, psi of x and t dx equal 1. You can imagine that not all kind of functions will satisfy it. In
  particular, any wave function, for example, that at infinity approaches a constant will never
  satisfy this, because if infinity, you approach a constant, then the integral is going to be infinite.
  And it’s just not going to work out. So the wave function cannot approach a finite number, a
  finite constant as x goes to infinity. So in order for this to hold-- order to guarantee this can
  even hold, can conceivably hold, it will require a bit of boundary conditions. And we’ll say that
  the limit as x goes to infinity or minus infinity-- plus/minus infinity of psi of x and t will be equal
  to 0. It better be true. And we’ll ask a little more.
  Now, you could say, look, certainly the limit of this function could not be in number, because it
  would be non-zero number, the interval will diverge. But maybe there is no limit. The wave
  function is so crazy that it can be integrated, but suddenly, it has a little spike and it just
  doesn’t have a normal limit. That could conceivably be the case. Nevertheless, it doesn’t seem
  to happen in any example that is of relevance. So we will assume that the situations are not
  that crazy that this happened.
  So we’ll take wave functions that necessarily go to 0 at infinity. And that certainly is good. You
  cannot prove it’s a necessary condition, but if it holds, it simplifies many, many things, and
  essentially, if the wave function is good enough to have a limit, then the limit must be 0.
  The other thing that we will want is that d psi/dx, the limit as x goes to plus/minus infinity is
  bounded. That is, yes, the limit may exist and it may be a number, but it’s not infinite. And In
  every example that I know of-- in fact, when this goes to 0, this goes to 0 as well-- but this is
  basically all you will ever need in order to make sense of the wave functions and their integrals
  that we’re going to be doing.
  Now you shouldn’t be too surprised that you need to say something about this wave function in
  the analysis that will follow, because the derivative-- you have the function and its derivative,
  because certainly, there are two derivatives here. So when we manipulate these quantities
  inside the integrals, you will see very soon-- single derivatives will show up and we’ll have to
  control them.
  So the only thing that I’m saying is that when you see a wave function that satisfies this
  property, you know that unless the function is extremely crazy, it’s a function that goes to 0 at
  plus/minus infinity. And it’s the relative pursuant it also goes to 0, but it will be enough to say
  that it maybe goes to a number.
  Now there’s another possibility thing for confusion here with things that we’ve been saying
  before. We’ve said before that the physics of a wave function is not altered by multiplying the
  wave function by a number. We said that psi added to psi is the same state; psi is the same
  state as square root of 2 psi-- all this is the same physics, but here it looks a little surprising if
  you wish, because if I have a psi and I got this already working out, if I multiply psi by square
  root of 2, it will not hold. So there seems to be a little maybe something with the words that
  we’re been using. It’s not exactly right and I want to make sure there is no room for confusion
  here, and it’s the following fact.
  Here, this wave function has been normalized. So there’s two kinds of wave functions that you
  can have-- wave functions that can be normalized and wave functions that cannot be
  normalized. Suppose somebody comes to you and gives you a psi of x and t. Or let’s assume
  that-- I’ll put x and t. No problem. Now suppose you go and start doing this integral-- integral of
  psi squared dx. And then you find that it’s not equal to 1 but is equal to some value N, which is
  different from 1 maybe. If this happens, we say that psi is normalizable, which means it can be
  normalized.
  And using this idea that changing the value-- the coefficient of the function-- doesn’t change
  too much, we simply say, use instead psi prime, which is equal to psi over square root of N.
  And look what a nice property this psi prime has. If you integrate psi prime squared, it would
  be equal-- because you have psi prime here is squared, it would be equal to the integral of PSI
  squared divided by the number N-- because there’s two of them-- dx, and the number goes
  out and you have the integral of psi squared dx, but that integral was exactly N, so that’s 1.
  So if your wave function has a finite integral in this sense, a number that is less than infinity,
  then psi can be normalized. And if you’re going to work with probabilities, you should use
  instead this wave function, which is the original wave function divided by a number. So they
  realize that, in some sense, you can delay all of this and you can always work with wave
  functions that are normalizable, but only when you’re going to calculate your probabilities. You
  can take the trouble to actually normalize them and those are the ones you use in these
  formulas.
  So the idea remains that we work flexibly with wave functions and multiply them by numbers
  and nothing changes as long as you realize that you cannot change the fact that the wave
  function is normalizable by multiplying it by any finite number, it will still be normalized. And if
  it’s normalizable, it’s equivalent to a normalized wave function. So those two words sound very
  similar, but they’re a little different. One is normalizable, which means it has an integral of psi
  squared finite, and normalize is one that already has been adjusted to do this and can be used
  to define a probability distribution.
  OK. So that, in a way of introduction to the problem that we have to do, our serious problem is
  indeed justifying that the time evolution doesn’t mess up the normalization and how does it do
  that?

L6.2 Is probability conserved? Hermiticity of the Hamiltonian (20:42)

MITOCW | watch?v=5L4QfjbK87M
PROFESSOR: Let’s do a work check. So main check. If integral psi star x t0, psi x t0 dx is equal to 1 at t equal
to t0, as we say there, then it must hold for later times, t greater than t0. This is what we want
to check, or verify, or prove.
Now, to do it, we’re going to take our time. So it’s not going to happen in five minutes, not 10
minutes, maybe not even half an hour. Not because it’s so difficult. It’s because there’s so
many things that one can say in between that teach you a lot about quantum mechanics. So
we’re going to take our time here.
So we’re going to first rewrite it with better notation. So we’ll define rho of x and t, which is
going to be called the probability density. And it’s nothing else than what you would expect, psi
star of x and t, psi of x and t. It’s a probability density. You know that has the right
interpretation, it’s psi squared. And that’s the kind of thing that integrated over space gives you
the total probability. So this is a positive number given by this quantity is called the probability
density. Fine.
What do we know about this probability density that we’re trying to find about its integral? So
define next N of t to be the integral of rho of x and t dx. Integrate this probability density
throughout space, and that’s going to give you N of t.
Now, what do we know? We know that N of t, or let’s assume that N of t0 is equal to 1. N is
that normalization. It’s that total integral of the probability what had to be equal to 1. Well, let’s
assume N at t0 is equal to 1. That’s good. The question is, will the Schrodinger equation
guarantee that-- and here’s the claim-- dN dt is equal to 0? Will the Schrodinger equation
guarantee this?
If the Schrodinger equation guarantees that this derivative is, indeed, zero, then we’re in good
business. Because the derivative is zero, the value’s 1, will remain 1 forever. Yes?
AUDIENCE: May I ask why you specified for t greater than t0?
Well, I don’t have to specify for t greater than t naught. I could do it for all t different than t
naught. But if I say this way, as imagining that somebody prepares the system at some time, t
naught, and maybe the system didn’t exist for other times below. Now, if a system existed for
long time and you look at it at t naught, then certainly the Schrodinger equation should imply
that it works later and it works before. So it’s not really necessary, but no loss of generality.
OK, so that’s it. Will it guarantee that? Well, that’s our thing to do. So let’s begin the work by
doing a little bit of a calculation. And so what do we need to do? We need to find the derivative
of this quantity. So what is this derivative of N dN dt will be the integral d dt of rho of x and t dx.
So I went here and brought in the d dt, which became a partial derivative. Because this is just
a function of t, but inside here, there’s a function of t and a function of x. So I must make clear
that I’m just differentiating t.
So is d dt of rho. And now we can write it as integral dx. What this rho? Psi star psi. So we
would have d dt of psi star times psi plus psi star d dt of psi.
OK. And here you see, if you were waiting for that, that the Schrodinger equation has to be
necessary. Because we have the psi dt. And that information is there with Schrodinger’s
equation. So let’s do that.
So what do we have? ih bar d psi dt equal h psi. We’ll write it like that for the time being
without copying all what h is. That would take a lot of time. And from this equation, you can
find immediately that d psi dt is minus i over h bar h hat psi.
Now we need to complex conjugate this equation, and that is always a little more scary.
Actually, the way to do this in a way that you never get into scary or strange things. So let me
take the complex conjugate of this equation. Here I would have i goes to minus i h bar, and
now I would have-- we can go very slow-- d psi dt star equals, and then I’ll be simple minded
here. I think it’s the best. I’ll just start the right hand side. I start the left hand side and start the
right hand side.
Now here, the complex conjugate of a derivative, in this case I want to clarify what it is. It’s just
the derivative of the complex conjugate. So this is minus ih bar d/dt of psi star equals h hat psi
star, that’s fine. And from here, if I multiply again by i divided by h bar, we get d psi star dt is
equal to i over h star h hat psi star.
We obtain this useful formula and this useful formula, and both go into our calculation of dN dt.
So what do we have here? dN dt equals integral dx, and I will put an i over h bar, I think, here.
Yes. i over h bar. Look at this term first. We have i over h bar, h psi star psi. And the second
term involves a d psi dt that comes with an opposite sign. Same factor of i over h bar, so
minus psi star h psi.
So the virtue of what we’ve done so far is that it doesn’t look so bad yet. And looks relatively
clean, and it’s very suggestive, actually. So what’s happening? We want to show that dN dt is
equal to 0. Now, are we going to be able to show that simply that to do a lot of algebra and
say, oh, it’s 0? Well, it’s kind of going to work that way, but we’re going to do the work and
we’re going to get to dN dt being an integral of something. And it’s just not going to look like 0,
but it will be manipulated in such a way that you can argue it’s 0 using the boundary condition.
So it’s kind of interesting how it’s going to work. But here structurally, you see what must
happen for this calculation to succeed. So we need for this to be 0. We need the following
thing to happen. The integral of h hat psi star psi be equal to the integral of psi star h psi. And I
should write the dx’s. They are there.
So this would guarantee that dN dt is equal to 0. So that’s a very nice statement, and it’s kind
of nice is that you have one function starred, one function non-starred. The h is where the
function needs to be starred, but on the other side of the equation, the h is on the other side.
So you’ve kind of moved the h from the complex conjugated function to the non-complex
conjugated function. From the first function to this second function.
And that’s a very nice thing to demand of the Hamiltonian. So actually what seems to be
happening is that this conservation of probability will work if your Hamiltonian is good enough
to do something like this. And this is a nice formula, it’s a famous formula. This is true if H is a
Hermitian operator.
It’s a very interesting new name that shows up that an operator being Hermitian. So this is
what I was promising you, that we’re going to do this, and we’re going to be learning all kinds
of funny things as it happens. So what is it for a Hermitian operator? Well, a Hermitian
operator, H, would actually satisfy the following. That the integral, H psi 1 star psi 2 is equal to
the integral of psi 1 star H psi 2.
So an operator is said to be Hermitian if you can move it from the first part to the second part
in this sense, and with two different functions. So this should be possible to do if an operator is
to be called Hermitian. Now, of course, if it holds for two arbitrary functions, it holds when the
two functions are the same, in this case.
So what we need is a particular case of the condition of hermiticity. Hermiticity simply means
that the operator does this thing. Any two functions that you put here, this equality is true. Now
if you ask yourself, how do I even understand that? What allows me to move the H from one
side to the other? We’ll see it very soon. But it’s the fact that H has second derivatives, and
maybe you can integrate them by parts and move the derivatives from the psi 1 to the psi 2,
and do all kinds of things.
But you should try to think at this moment structurally, what kind of objects you have, what kind
of properties you have. And the objects are this operator that controls the time evolution,
called the Hamiltonian. And if I want probability interpretation to make sense, we need this
equality, which is a consequence of hermiticity.
Now, I’ll maybe use a little of this blackboard. I haven’t used it much before. In terms of
Hermitian operators, I’m almost there with a definition of a Hermitian operator. I haven’t quite
given it to you, but let’s let state it, given that we’re already in this discussion of hermiticity.
So this is what is called the Hermitian operator, does that. But in general, rho, given an
operator T, one defines its hermitian conjugate P dagger as follows. So you have the integral
of psi 1 star T psi 2, and that must be rearranged until it looks like T dagger psi 1 star psi 2.
Now, these things are the beginning of a whole set of ideas that are terribly important in
quantum mechanics. Hermitian operators, or eigenvalues and eigenvectors. So it’s going to
take a little time for you to get accustomed to them. But this is the beginning. You will explore a
little bit of these things in future homework, and start getting familiar. For now, it looks very
strange and unmotivated. Maybe you will see that that will change soon, even throughout
today’s lecture.
So this is the Hermitian conjugate. So if you want to calculate the Hermitian conjugate, you
must start with this thing, and start doing manipulations to clean up the psi 2, have nothing at
the psi 2, everything acting on psi 1, and that thing is called the dagger.
And then finally, T is Hermitian if T dagger is equal to T. So its Hermitian conjugate is itself. It’s
almost like people say a real number is a number whose complex conjugate is equal to itself.
So a Hermitian operator is one whose Hermitian conjugate is equal to itself, and you see if T is
Hermitian, well then it’s back to T and T in both places, which is what we’ve been saying here.
This is a Hermitian operator.

L6.3 Probability current and current conservation (15:14)

MITOCW | watch?v=J2ltXyByPJA
BARTON
ZWIEBACH:
After this long detour, you must think that one is just trying to avoid doing the real computation,
so here comes, the real computation. The real computation is taking that right hand side on
the top of the blackboard and trying to just calculate this right hand side. So back to the
calculation. The calculation dN/dt is equal to this thing over there, integral dx i over h-bar. I’ll
still copy it here-- h psi-star psi minus psi-star h psi.
OK. Well, let’s do this. This whole quantity is d-rho/dt, and let’s see how much it is. Well, you
would have of the following-- i over h-bar h psi-star. Well, h in detail is over there, so I’ll put it
here. Minus h squared over 2m d second dx squared of psi-star. So I’m beginning h psi star–
that’s from the first term in the Hamiltonian-- times psi. And then from the other term in the
Hamiltonian is the potential, so it would be plus V of x and t psi-star psi. V of x and t times psistar times psi.
This other term would be minus psi-star h psi, so it’s going to be opposite sign to here, so plus
h squared over 2m psi-star d second dx squared psi, and then minus psi-star V of x and t psi.
Here, there was a little thing that I probably should have said before is that the potential is real,
that’s why it didn’t get complex conjugated here. H psi would have a term V psi and we just
conjugate the psi.
OK, this is not so bad. In particular, you see that these two terms cancel. So that’s neat. And
now, this becomes the following-- this d-rho/dt has become minus ih over 2m d second psistar dx squared times psi minus psi-star d second psi dx squared. OK. That’s what d-rho/dt is
and that’s the thing that should be 0 when you integrate-- it doesn’t look like anything equal to
0, and that was pretty much to be expected.
So what do we have to do with this? Well, we have to simplify it more, and what could save us
is, and it’s usually the same thing that saves you all the time when you want to show an
integral vanishes, many times, what you show is that it is a total derivative. So remember,
we’re computing here d-rho/dt, which is all this thing circled here, and it’s to be integrated over
x. So if I could show this is a derivative with respect to x, the total x derivative, then the integral
would go to the boundaries and I would have a chance to make it 0.
So what do we have? That derivative is indeed at boundaries, so d-rho/dt is equal to minus i hbar bar over 2m. And look, this can be written as d/dx of something and what is that
something? It’s d psi-star dx times psi minus psi-star d second psi-- no not d second-- d first
psi dx. The nice thing that happens here is that if you act with this d/dx, you get the second
derivative terms that you had in there. But you also get derivatives acting here on d psi and
here on d psi-star, but those will cancel. So it’s a very lucky circumstance, it had better
happen, but this is a total derivative with respect to x. And that’s just very a good deal.
So we’re going to rewrite it a little more. I’ll write it as the following way-- this whole factor is h
over 2im, that’s with its sign, output the d/dx outside-- I’ll put an extra minus sign, so I will flip
the order of these two terms-- psi-star d psi dx minus psi d psi-star dx. OK.
Well, in many ways, the most difficult part of the calculation is over and it’s now a matter of
giving proper names to things. Why do I say that? Because look, want to see the finish line?
It’s here. We’ve shown this whole integrand is d/dx of that right hand side. Therefore, when
you do the integral, you will have to go to the boundary with that thing, so you just need to see
what happens to these quantities as x goes to infinity. And as x goes to infinity, we said that psi
must go to 0 from the beginning. And d psi dx must not blow up, so if psi goes to 0 and d psi
dx doesn’t blow up, this whole thing goes to 0 and dN/dt is equal to 0 and you’re done. So
you’re done with the conditions that we mention that the wave function must satisfy these
conditions.
But let’s clean up this, because we’ve actually discovered an important quantity over there that
is going to play a role. So here you see that you have a complex number minus its complex
conjugate. So this is like z minus z-star, which is equal to 2i I times the imaginary part of z. If
you subtract from a complex number its complex conjugate, you get the imaginary part only
survives, but it’s twice of it. So from here, this whole thing is 2i times the imaginary part of psistar d psi dx.
So d-rho/dt is equal to minus d/dx of what? Of 2i times the imaginary part of that, cancels the
2i, you get h-bar over m imaginary part of psi-star d psi dx. And this quantity is going to be
called the current density. So the current density, you say, why the current density? We’ll see
in a minute. But let’s write it here because it’ll be very important. J of x and t is h-bar over m
imaginary part of psi-star d psi dx.
So if this is called the current density, you would have an equation here d-rho/dt is equal to
minus dJ/dx d/dx of J dx, or d-rho/dt plus dJ/dx is equal to 0. Now this is called current
conservation. You’ve seen it before in electromagnetism and we’ll review it here in a second
as well.
So look what has happened. You began with the introduction of a charged density, which was
a probability density, but you were led now to the existence of a current. And you’ve seen that
in three dimensions, more than in one dimension-- I think probably in one dimension it doesn’t
look that familiar to you, but let me make sure you will recognize it in a few seconds.
So think units here first. Units. What are the units of the wave function? Well, the wave
function, you integrate over x squared and it gives you 1. So the integral of psi squared dx is
equal to 1, so this has units of length, this must have units of 1 over square root of length. And
what are therefore the units of psi-star d/dx psi, which is part of the current formula ? Well, 1 of
the square root of length-- 1 over square root of length is one over length and another 1 over
length is 1 over length squared. OK.
And then you have h-bar, which has units of mL squared over T. Probably done that before
already. And therefore, h over m has units of L squared over T. So the current has units of h
over m-- the units of current has units of h over m, which is L squared over T-- times units of
this whole thing, which is 1 over L squared, so at the end, 1 over T. And this means just
probability per unit time. That’s the units of current. Probability has no units, so we’re dealing
probability, those are pure numbers, but this is probability per unit time. So probability per unit
time.

L6.4 Three dimensional current and conservation (18:13)

MITOCW | watch?v=Ex_fFlwZoM0
Three-dimensional case. Now, in the future homework, you will be doing the equivalent of this calculation here
with the Laplacians-- it’s not complicated-- so that you will derive with the current is.
And the current must be a very similar formula as this one. And indeed, I’ll just write it here. The current is h bar
over m, the imaginary part of psi star. And instead of ddx, you expect the gradient of psi. That is the current for
the probability in three dimensions.
And the analog of this equation, d rho dt plus dj dx equals 0, is d rho dt plus divergence of j is equal to 0. That is
current conservation. Perhaps you do remember that from your study of electromagnetism. That’s how Maxwell
discovered the displacement current when he tried to figure out how everything was compatible with current
conservation. Anyway, that argument I’ll do in a second so that it will become clearer.
So one last thing here-- it’s something also-- you can check the units here of j is 1 over l squared times 1 over t,
so probability per unit area and unit time.
So what did we have? We were doing the integral of the derivative of the integral given by n. It was over here, dn
dt. We worked hard on it. And dn dt was the integral of d rho dt. So it was the integral of d rho dt dx.
But we showed now that d rho dt is minus dj dx. So here you have integral from minus infinity to infinity dx of dj dx.
And therefore, this is-- I should have a minus sign, because it was minus dj dx. This is minus the current of x
equals infinity times p minus the current at x equals minus infinity nt.
And as we more or less hinted before, since the current is equal to h over 2im psi star [INAUDIBLE] psi dx minus
psi [INAUDIBLE] psi star dx, as you go to plus infinity or minus infinity, these things go to 0 given the boundary
conditions that we put. Because psi or psi star go to 0 to infinity, and the derivates are bound at the infinity.
So this is 0, dn dt 0. All is good. And two things happened. In the way of doing this, we realized that the
computation we have done pretty much established that this is equal to that, because dn dt is the difference of
these two integrals, and we showed it’s 0. So this is true.
And therefore, we suspect h is a Hermitian operator. And the thing that we should do in order to make sure it is is
put two different functions here, not two equal functions. This worked for two equal function, but for two different
functions, and check it as well. And we’ll leave it as an exercise. It’s a good exercise.
So this shows the consistency. But we discover two important ideas-- one, the existence of a current for
probability, and two, h is a Hermitian operator.
So last thing is to explain the analogy with current conservation. I think this should help as well. So the
interpretation that we’ll have is the same as we have in electromagnetism. And there’s a complete analogy for
everything here. So not for the wave function, but for all these charge densities and current densities.
So we have electromagnetism and quantum mechanics. We have rho. Here is the charge density. And here is the
probability density. If you have a total charge q in a volume, here is the probability to find the particle in a volume.
There is a j in Maxwell’s equations as well, and that’s a current density. Amber’s law has that current. It generates
the curl of b. And here is a probability current density. So that’s the table.
So what I want to make sure is that you understand why these equations, like this or that, are more powerful than
just showing that dn dt is 0. They imply a local conservation of probability. You see, there has to be physics of this
thing. So the total probability must be 1.
But suppose you have the probability distributed over space. There must be some relation between the way the
probabilities are varying at one point and varying in other points so that everything is consistent. And those are
these differential relations that say that whenever you see a probability density change anywhere, it’s because
there is some current.
And that makes sense. If you see the charge density in some point in space changing, it’s because there must be
a current. So thanks to the current, you can learn how to interpret the probability much more physically. Because if
you ask what is the probability that the particle is from this distance to that distance, you can look at what the
currents are at the edges and see whether that probability is increasing or decreasing.
So let’s see that. Suppose you have a volume, and define the charge inside the volume. Then you say OK, does
this charge change in time? Sure, it could. So dq dt is equal to integral d rho dt d cube x over the volume.
But d rho dt, by the current conservation equation-- that’s the equation we’re trying to make sure your intuition is
clear about-- this is equal to minus the integral of j-- no, of divergence of j d cube x over the volume.
OK. But then Gauss’s law. Gauss’s Lot tells you that you can relate this divergence to a surface integral. dq dt is
therefore minus the surface integral, the area of the current times that. So I’ll write it as minus jda, the flux of the
current, over the surface that bounds-- this is the volume, and there’s the surface bounding it.
So by the divergence theorem, it becomes this. And this is how you understand current conservation. You say,
look, charge is never created or destroyed. So if you see the charge inside the volume changing, it’s because
there’s some current escaping through the surface.
So that’s the physical interpretation of that differential equation, of that d rho dt plus divergence of j is equal to 0.
This is current conservation. And many people look at this equation and say, what? Current conservation? I don’t
see anything.
But when you look at this equation, you say, oh, yes. The charge changes only because it escapes the volume,
not created nor destroyed. So the same thing happens for the probability.
Now, let me close up with this statement in one dimension, which is the one you care, at this moment. And on the
line, you would have points a and b. And you would say the probability to be within a and b is the integral from a to
b dx of rho. That’s your probability. That’s the integral of psi squared from a to b.
Now, what is the time derivative of it? dp ab dt would be integral from a to b dx of d rho dt. But again, for that
case, d rho dt is minus dj dx. So this is minus dx dj dx between b and a.
And what is that? Well, you get the j at the boundary. So this is minus j at x equals bt minus j at x equal a, t. So
simplifying it, you get dp ab dt is equal to minus j at x equals bt plus j of a, t.
Let’s see if that makes sense. You have been looking for the particle and decided to look at this range from a to b.
That’s the probability to find it there. We learned already that the total probability to find it anywhere is going to be
1, and that’s going to be conserved, and it’s going to be no problems.
But now let’s just ask given what happens to this probability in time. Well, it could change, because the wave
function could go up and down. Maybe the wave function was big here and a little later is small so there’s less
probability to find it here.
But now you have another physical variable to help you understand it, and that’s the current. That formula we
found there for j of x and t in the upper blackboard box formula is a current that can be computed. And here you
see if the probability to find the particle in this region changes, it’s because some current must be escaping from
the edges.
And let’s see if the formula gives it right. Well, we’re assuming quantities are positive if they have plus components
in the direction of x. So this current is the current component in the x direction. And it should not be lost-- maybe I
didn’t quite say it-- that if you are dealing with a divergence of j, this is dj x dx plus dj y dy plus dj z dz. And in the
case of one dimension, you will have those, and you get this equation. So it’s certainly the reduction.
But here you see indeed, if the currents are positive-- if the current at b is positive, there is a current going out. So
that tends to reduce the probability. That’s why the sign came out with a minus.
On the other hand, if there is a current in a, that tends to send in probability, and that’s why it increases it here. So
the difference between these two currents determines whether the probability here increases or decreases.