WEBVTT

00:00:00.500 --> 00:00:03.110
PROFESSOR: This
definition in which

00:00:03.110 --> 00:00:06.892
the uncertainty of the
permission operator

00:00:06.892 --> 00:00:10.860
Q in the state psi.

00:00:10.860 --> 00:00:14.150
It's always important to
have a state associated

00:00:14.150 --> 00:00:16.530
with measuring the uncertainty.

00:00:16.530 --> 00:00:19.760
Because the uncertainty will be
different in different states.

00:00:19.760 --> 00:00:22.790
So the state should
always be there.

00:00:22.790 --> 00:00:24.580
Sometimes we write
it, sometimes we

00:00:24.580 --> 00:00:28.183
get a little tired of writing
it and we don't write it.

00:00:28.183 --> 00:00:31.890
But it's always implicit.

00:00:31.890 --> 00:00:33.840
So here it is.

00:00:33.840 --> 00:00:37.650
From the analogous discussion
of random variables,

00:00:37.650 --> 00:00:40.190
we were led to this
definition, in which we

00:00:40.190 --> 00:00:42.260
would have the expectation
value of the square

00:00:42.260 --> 00:00:46.862
of the operator minus the
square of the expectation value.

00:00:46.862 --> 00:00:53.820
This was always-- well, this
is always a positive quantity.

00:00:53.820 --> 00:01:01.050
Because, as claim 1 goes, it can
be rewritten as the expectation

00:01:01.050 --> 00:01:07.964
value of the square of the
difference between the operator

00:01:07.964 --> 00:01:10.490
and its expectation value.

00:01:10.490 --> 00:01:13.050
This may seem a little strange.

00:01:13.050 --> 00:01:16.230
You're subtracting from
an operator a number,

00:01:16.230 --> 00:01:22.806
but we know that numbers can be
thought as operators as well.

00:01:22.806 --> 00:01:28.305
Operator of minus a
number acting on a state

00:01:28.305 --> 00:01:29.440
is well defined.

00:01:29.440 --> 00:01:32.780
The operator acts on the state,
the number multiplies a state.

00:01:32.780 --> 00:01:34.533
So this is well defined.

00:01:34.533 --> 00:01:39.050
And claim 1 is proven
by direct computation.

00:01:39.050 --> 00:01:44.400
You certainly indeed prove.

00:01:44.400 --> 00:01:48.070
You can expand what is
inside the expectation value,

00:01:48.070 --> 00:01:52.000
so it's Q hat squared.

00:01:52.000 --> 00:01:58.270
And then the double product
of this Q hat and this number.

00:01:58.270 --> 00:02:01.560
Now, the number
and Q hat commute,

00:02:01.560 --> 00:02:04.280
so it is really
the double product.

00:02:04.280 --> 00:02:09.595
If you have A plus B times A
plus B, you have AB plus BA,

00:02:09.595 --> 00:02:12.120
but if they commute
it's 2AB, so this

00:02:12.120 --> 00:02:20.488
is minus 2 Q hat Q. Like that.

00:02:20.488 --> 00:02:24.100
And then, the last term
is the number squared,

00:02:24.100 --> 00:02:28.856
so it's plus Q squared.

00:02:28.856 --> 00:02:34.235
And sometimes I don't
put the hats as well.

00:02:34.235 --> 00:02:37.020
And all this is the
expectation value

00:02:37.020 --> 00:02:39.964
of the sum of all these things.

00:02:39.964 --> 00:02:43.130
The expectation value
of a sum of things

00:02:43.130 --> 00:02:44.882
is the expectation
value of the first

00:02:44.882 --> 00:02:46.590
plus the expectation
value of the second,

00:02:46.590 --> 00:02:49.170
plus the expectation
value of the next.

00:02:49.170 --> 00:02:54.586
So we can go ahead and do
this, and this is therefore

00:02:54.586 --> 00:03:00.696
expectation value of Q
squared minus the expectation

00:03:00.696 --> 00:03:02.620
value of this whole thing.

00:03:05.260 --> 00:03:10.386
But now the expectation value
of a number times an operator,

00:03:10.386 --> 00:03:11.730
the number can go out.

00:03:11.730 --> 00:03:14.185
And this is a number,
and this is a number.

00:03:14.185 --> 00:03:21.292
So it's minus 2 expectation
value of Q, number went out.

00:03:21.292 --> 00:03:29.148
And then you're left with
expectation value of another Q.

00:03:29.148 --> 00:03:33.282
And the expectation
value of a number

00:03:33.282 --> 00:03:35.782
is just the number,
because then you're

00:03:35.782 --> 00:03:39.930
left within the world of psi
star psi, which is equal to 1.

00:03:39.930 --> 00:03:48.700
So here is plus Q hat squared.

00:03:48.700 --> 00:03:52.880
And these two terms, the
second and the third,

00:03:52.880 --> 00:03:54.700
are the same really.

00:03:54.700 --> 00:03:59.090
They are both equal to
expectation value of Q squared.

00:03:59.090 --> 00:04:03.212
They cancel a little bit,
and they give you this.

00:04:03.212 --> 00:04:08.280
So indeed, this is equal
to expectation value

00:04:08.280 --> 00:04:13.293
of Q squared minus expectation
value of Q squared.

00:04:13.293 --> 00:04:16.290
So claim 1 is true.

00:04:16.290 --> 00:04:22.089
And claim 1 shows in particular
that this number, delta Q

00:04:22.089 --> 00:04:28.180
squared, in the expectation
value of a square of something,

00:04:28.180 --> 00:04:29.450
is positive.

00:04:29.450 --> 00:04:32.910
We'll see more
clearly in a second

00:04:32.910 --> 00:04:38.156
when we have claim number 2.

00:04:38.156 --> 00:04:42.638
And claim number 2
is easily proven.

00:04:42.638 --> 00:04:48.660
That's another expression
for uncertainty.

00:04:48.660 --> 00:04:57.350
For claim number 2, we will
start with the expectation

00:04:57.350 --> 00:05:08.298
value of Q minus Q squared, like
this, which is the integral dx

00:05:08.298 --> 00:05:19.676
psi star of x and t, Q minus
expectation value of Q, Q

00:05:19.676 --> 00:05:23.787
minus expectation
value of Q, on psi.

00:05:32.860 --> 00:05:36.010
The expectation value
of this thing squared

00:05:36.010 --> 00:05:39.614
is psi star, the
operator, and this.

00:05:39.614 --> 00:05:49.503
And now, think of this as an
operator acting on all of that.

00:05:49.503 --> 00:05:51.938
This is a Hermitian operator.

00:06:03.650 --> 00:06:15.550
Because Q hat is Hermitian, and
expectation value of Q is real.

00:06:15.550 --> 00:06:19.710
So actually this real
number multiplying something

00:06:19.710 --> 00:06:23.200
can be moved from
the wave function

00:06:23.200 --> 00:06:26.830
to the starred wave
function without any cost.

00:06:26.830 --> 00:06:31.145
So even though you might
not think of a real number

00:06:31.145 --> 00:06:35.386
as a Hermitian operator, it is.

00:06:35.386 --> 00:06:39.370
And therefore this whole
thing is Hermitian.

00:06:39.370 --> 00:06:42.730
So it can be written as dx.

00:06:42.730 --> 00:06:47.206
And now you have this
whole operator, Q minus Q

00:06:47.206 --> 00:06:52.695
hat, acting on psi of x and t.

00:06:52.695 --> 00:06:56.480
And conjugate.

00:06:56.480 --> 00:06:59.784
Remember, the operator,
the Hermitian operator,

00:06:59.784 --> 00:07:04.924
moves to act on psi, and
the whole thing [INAUDIBLE].

00:07:04.924 --> 00:07:10.682
And then we have here
the other term left over.

00:07:14.948 --> 00:07:21.940
But now, you see that you
have whatever that state is

00:07:21.940 --> 00:07:24.438
and the state
complex conjugated.

00:07:27.860 --> 00:07:31.083
So that is equal
to this integral.

00:07:31.083 --> 00:07:38.472
This is the integral dx of the
norm squared of Q hat minus Q

00:07:38.472 --> 00:07:46.785
hat psi of x and t squared,
which means that thing, that's

00:07:46.785 --> 00:07:50.430
its complex conjugate.

00:07:50.430 --> 00:07:55.315
So this completes
our verification

00:07:55.315 --> 00:07:58.780
that these claims are
true, and allow us

00:07:58.780 --> 00:08:04.270
to do the last step on
this analysis, which

00:08:04.270 --> 00:08:12.450
is to show that if you
have an eigenstate of Q,

00:08:12.450 --> 00:08:18.190
if a state psi is an eigenstate
of Q, there is no uncertainty.

00:08:18.190 --> 00:08:21.586
This goes along with our
measurement postulate that

00:08:21.586 --> 00:08:29.270
says an eigenstate
of Q, you measure Q

00:08:29.270 --> 00:08:34.653
and you get the eigenvalue of
Q and there's no uncertainty.

00:08:34.653 --> 00:08:37.480
In particular, we'll
do it here I think.

00:08:37.480 --> 00:08:52.974
If psi is an eigenstate
of Q, so you'll

00:08:52.974 --> 00:09:04.790
have Q psi equal lambda psi,
where lambda is the eigenvalue.

00:09:04.790 --> 00:09:11.580
Now, this is a nice thing.

00:09:11.580 --> 00:09:17.550
It's stating that the state
psi is an eigenstate of Q

00:09:17.550 --> 00:09:21.505
and this is the eigenvalue,
but there is a little bit more

00:09:21.505 --> 00:09:23.273
than can be said.

00:09:23.273 --> 00:09:26.280
And it is.

00:09:26.280 --> 00:09:29.480
It should not surprise
you that the eigenvalue

00:09:29.480 --> 00:09:36.720
happens to be the expectation
value of Q on the state psi.

00:09:36.720 --> 00:09:37.500
Why?

00:09:37.500 --> 00:09:42.310
Because you can take this
equation and integrate dx times

00:09:42.310 --> 00:09:43.710
psi star.

00:09:43.710 --> 00:09:47.120
If you bring that in into
both sides of the equation

00:09:47.120 --> 00:09:58.984
then you have Q psi equals
integral dx psi star psi,

00:09:58.984 --> 00:10:02.900
and the lambda goes up.

00:10:02.900 --> 00:10:08.260
Since my assumption whenever
you do expectation values,

00:10:08.260 --> 00:10:12.990
your states are normalized,
this is just lambda.

00:10:12.990 --> 00:10:21.470
And by definition, this is
the expectation value of Q.

00:10:21.470 --> 00:10:26.130
So lambda happens to be equal
to the expectation value of Q,

00:10:26.130 --> 00:10:31.280
so sometimes we can say
that this equation really

00:10:31.280 --> 00:10:40.490
implies that Q hat psi is equal
to expectation value of Q psi

00:10:40.490 --> 00:10:41.957
times psi.

00:10:53.210 --> 00:10:55.460
It looks a little
strange in this form.

00:10:55.460 --> 00:10:57.530
Very few people write
it in this form,

00:10:57.530 --> 00:11:01.580
but it's important to recognize
that the eigenvalue is

00:11:01.580 --> 00:11:03.700
nothing else but the
expectation value

00:11:03.700 --> 00:11:06.486
of the operator of that state.

00:11:06.486 --> 00:11:10.620
But if you recognize
that, you realize

00:11:10.620 --> 00:11:21.945
that the state satisfies
precisely Q hat minus Q on psi

00:11:21.945 --> 00:11:25.670
is equal to 0.

00:11:25.670 --> 00:11:31.470
Therefore, if Q hat minus
Q on psi is equal to 0,

00:11:31.470 --> 00:11:33.684
delta Q is equal to 0.

00:11:38.960 --> 00:11:42.800
By claim 2.

00:11:45.460 --> 00:11:50.390
Q hat minus Q expectation
value kills the state,

00:11:50.390 --> 00:11:52.954
and therefore this is 0.

00:11:56.300 --> 00:11:57.880
OK then.

00:11:57.880 --> 00:11:59.840
The other way is also true.

00:11:59.840 --> 00:12:12.210
If delta Q is equal to 0, by
claim 2, this integral is 0.

00:12:12.210 --> 00:12:15.310
And since it's
the sum of squares

00:12:15.310 --> 00:12:22.882
that are always positive, this
state must be 0 by claim 2.

00:12:27.072 --> 00:12:35.866
And you get that Q minus
Q hat psi is equal to 0.

00:12:35.866 --> 00:12:46.700
And this means that psi
is an eigenstate of Q.

00:12:46.700 --> 00:12:49.250
So the other way
around it also works.

00:12:49.250 --> 00:12:54.500
So the final
conclusion is delta Q

00:12:54.500 --> 00:12:59.930
is equal to 0 is
completely equivalent of--

00:12:59.930 --> 00:13:02.460
I'll put in the psi.

00:13:02.460 --> 00:13:16.720
Psi is an eigenstate of Q.

00:13:16.720 --> 00:13:20.730
So this is the main conclusion.

00:13:27.440 --> 00:13:32.056
Also, we learned some
computational tricks.

00:13:32.056 --> 00:13:34.945
Remember you have to
compute an expectation

00:13:34.945 --> 00:13:38.750
value of a number,
uncertainty, you

00:13:38.750 --> 00:13:43.020
have these various
formulas you can use.

00:13:43.020 --> 00:13:45.770
You could use the
first definition.

00:13:45.770 --> 00:13:48.150
Sometimes it may
be the simplest.

00:13:48.150 --> 00:13:52.420
In particular, if the
expectation value of Q

00:13:52.420 --> 00:13:56.190
is simple, it's the easiest way.

00:13:56.190 --> 00:14:00.344
So for example, you can have
a Gaussian wave function,

00:14:00.344 --> 00:14:05.311
and people ask you, what is
delta of x of the Gaussian wave

00:14:05.311 --> 00:14:05.810
function?

00:14:08.760 --> 00:14:11.980
Well, on this Gaussian
wave function,

00:14:11.980 --> 00:14:14.060
you could say that
delta x squared

00:14:14.060 --> 00:14:17.760
is the expectation value of x
squared minus the expectation

00:14:17.760 --> 00:14:19.414
value of x squared.

00:14:22.080 --> 00:14:25.140
What is the
expectation value of x?

00:14:25.140 --> 00:14:29.150
Well, it would seem reasonable
that the expectation value of x

00:14:29.150 --> 00:14:29.833
is 0.

00:14:29.833 --> 00:14:32.560
It's a Gaussian
centered at the origin.

00:14:32.560 --> 00:14:33.960
And it's true.

00:14:33.960 --> 00:14:38.015
For a Gaussian it would be 0,
the expectation value of x.

00:14:38.015 --> 00:14:40.420
So this term is 0.

00:14:40.420 --> 00:14:43.140
You can also see 0
because of the integral.

00:14:43.140 --> 00:14:47.530
You're integrating x
against psi squared.

00:14:47.530 --> 00:14:51.890
Psi squared is even,
x is odd with respect

00:14:51.890 --> 00:14:54.880
to x going to minus x.

00:14:54.880 --> 00:14:57.686
So that integral
is going to be 0.

00:14:57.686 --> 00:15:02.140
So in this case, the
uncertainty is just

00:15:02.140 --> 00:15:04.920
the calculation of the
expectation value of x squared,

00:15:04.920 --> 00:15:06.952
and that's easily done.

00:15:06.952 --> 00:15:10.600
It's a Gaussian integral.

00:15:10.600 --> 00:15:14.230
The other good thing
about this is that

00:15:14.230 --> 00:15:17.033
even though we have not
proven the uncertainty

00:15:17.033 --> 00:15:20.265
principle in all generality.

00:15:20.265 --> 00:15:22.740
We've only [? multivated ?] it.

00:15:22.740 --> 00:15:25.400
It's precise with
this definition.

00:15:25.400 --> 00:15:29.940
So when you have
the delta x, delta p

00:15:29.940 --> 00:15:34.570
is greater than or
equal to h bar over 2,

00:15:34.570 --> 00:15:38.868
these things are computed
with those definitions.

00:15:38.868 --> 00:15:42.000
And then it's precise.

00:15:42.000 --> 00:15:44.490
It's a mathematically
rigorous result.

00:15:44.490 --> 00:15:46.800
It's not just hand waving.

00:15:46.800 --> 00:15:49.530
The hand waving is good.

00:15:49.530 --> 00:15:53.030
But the precise result
is more powerful.