# A query regarding Rotational Invariance

• I
We know that Bell States follow the Rotational Invariance property i.e. the probability of results on measurement of two entangled particles do not change if the initial measurement basis (say ##u##) is rotated by an angle θ to a new basis (to say ##v##).

Lets take the Bell State ##\psi = \frac 1 {\sqrt 2}(\uparrow \uparrow + \downarrow \downarrow)##. Here both the terms have the probability ##50##% in basis ##u## or rotated basis ##v##.

How does the Rotational Invariance property get affected if the probability of two terms is unequal i.e. if the entangled state is of the form ##\psi = (α\uparrow \uparrow + β\downarrow \downarrow)##, ##α^2+β^2=1##, ##α≠β##?

My guess is they remain unchanged as it is a generic case of Bell States.

Last edited:

Nugatory
Mentor
How well does your guess work for the extreme case ##\alpha=0## and ##\beta=1##?

How well does your guess work for the extreme case ##\alpha=0## and ##\beta=1##?

haha. That is why it was a guess. I stated that assuming non zero real values. I am not sure how to compute the generic case hence the Q?

PeroK
Homework Helper
Gold Member
2020 Award
We know that Bell States follow the Rotational Invariance property i.e. the probability of results on measurement of two entangled particles do not change if the initial measurement basis (say ##u##) is rotated by an angle θ to a new basis (to say ##v##).

Lets take the Bell State ##\psi = \frac 1 {\sqrt 2}(\uparrow \uparrow + \downarrow \downarrow)##. Here both the terms have the probability ##50##% in basis ##u## or rotated basis ##v##.

How does the Rotational Invariance property get affected if the probability of two terms is unequal i.e. if the entangled state is of the form ##\psi = (α\uparrow \uparrow + β\downarrow \downarrow)##, ##α^2+β^2=1##, ##α≠β##?

My guess is they remain unchanged as it is a generic case of Bell States.
You shouldn't be guessing, you should be calculating! I make it that the probability of getting spin up in a direction making angle ##\theta## with the z-axis is:
$$\beta^2 + (1 - 2\beta^2)\cos^2(\frac \theta 2)$$
You might want to double check that. It works out for ##\beta = \frac 1 {\sqrt 2}##.

vanhees71 and Student149
PeroK
Homework Helper
Gold Member
2020 Award
haha. That is why it was a guess. I stated that assuming non zero real values. I am not sure how to compute the generic case hence the Q?
I can post the details if you are interested.

I can post the details if you are interested.

Thank you. I would be thankful..

PeroK
Homework Helper
Gold Member
2020 Award
Thank you. I would be thankful..
The key is to work out the ##\uparrow_z## and ##\downarrow_z## states in an arbitrary basis for a measurement direction defined by spherical angles ##\theta, \phi## and denoted by ##n##. These are:
$$\uparrow_z = \cos(\frac \theta 2)\uparrow_n - \sin(\frac \theta 2)\downarrow_n$$ and
$$\downarrow_z = e^{-i\phi}(\sin(\frac \theta 2)\uparrow_n + \cos(\frac \theta 2)\downarrow_n)$$
You can plug these in to the required Bell state and expand. For the state you used in the OP, I get:
$$\psi = [\alpha \cos^2(\frac \theta 2) + \beta e^{-2i\phi}\sin^2(\frac \theta 2)]\uparrow_n\uparrow_n + [\alpha \sin^2(\frac \theta 2) + \beta e^{-2i\phi}\cos^2(\frac \theta 2)]\downarrow_n\downarrow_n + \\ \frac 1 2\sin\theta(\beta e^{-2i\phi} - \alpha)(\uparrow_n\downarrow_n + \downarrow_n\uparrow_n)$$
It would be good if you could double check that, as well as the probability calculations from that

vanhees71 and Student149
You shouldn't be guessing, you should be calculating! I make it that the probability of getting spin up in a direction making angle θ with the z-axis is:
β2+(1−2β2)cos2⁡(θ2)
You might want to double check that. It works out for β=12.
The key is to work out the ##\uparrow_z## and ##\downarrow_z## states in an arbitrary basis for a measurement direction defined by spherical angles ##\theta, \phi## and denoted by ##n##. These are:
$$\uparrow_z = \cos(\frac \theta 2)\uparrow_n - \sin(\frac \theta 2)\downarrow_n$$ and
$$\downarrow_z = e^{-i\phi}(\sin(\frac \theta 2)\uparrow_n + \cos(\frac \theta 2)\downarrow_n)$$
You can plug these in to the required Bell state and expand. For the state you used in the OP, I get:
$$\psi = [\alpha \cos^2(\frac \theta 2) + \beta e^{-2i\phi}\sin^2(\frac \theta 2)]\uparrow_n\uparrow_n + [\alpha \sin^2(\frac \theta 2) + \beta e^{-2i\phi}\cos^2(\frac \theta 2)]\downarrow_n\downarrow_n + \\ \frac 1 2\sin\theta(\beta e^{-2i\phi} - \alpha)(\uparrow_n\downarrow_n + \downarrow_n\uparrow_n)$$
It would be good if you could double check that, as well as the probability calculations from that

I would. much appreciated.

PeroK
Homework Helper
Gold Member
2020 Award
I would. much appreciated.
It all seems to check out. I can simplify the equations a bit:
$$p(\uparrow_n | \psi) = \frac 1 2 + (\frac 1 2 - |\beta|^2)\cos \theta = \frac 1 2 - (\frac 1 2 - |\alpha|^2)\cos \theta$$ $$p(\downarrow_n | \psi) = \frac 1 2 + (\frac 1 2 - |\alpha|^2)\cos \theta = \frac 1 2 - (\frac 1 2 - |\beta|^2)\cos \theta$$
Note that unless ##|\alpha^2| = |\beta^2| = \frac 1 2## these probabilities vary with ##\theta##. Note also that if ##\theta = \frac \pi 2## (i.e. we are measuring in the x-y plane), then the probabilities are ##1/2## independent of ##\alpha, \beta##.

Student149
PeroK
Homework Helper
Gold Member
2020 Award
PS physically we should have expected the probabilities to depend on ##\theta## but not on ##\phi##. I could have set ##\phi = 0## and made things easier.

vanhees71
Gold Member
Why don't the probs. add to 1? Also the very idea in #1 seems to be flawed to me. This is a state in the triplet, i.e., for ##S=1## and thus not rotationally invariant.

Why don't the probs. add to 1? Also the very idea in #1 seems to be flawed to me. This is a state in the triplet, i.e., for ##S=1## and thus not rotationally invariant.

I am confused. I think they do add to 1 in #9. Please elaborate what is flawed in #1 (I am a novice in QM, but I can't find a flaw)

vanhees71
Gold Member
The two-spin state in #1 is
$$|\Psi \rangle=\frac{1}{\sqrt{2}} (|1/2,1/2 \rangle + |-1/2,-1/2 \rangle).$$
This is a state belonging to total spin ##S=1##, i.e., the spin-triplet state and thus is not rotationally invariant.

I also do not understand the notation in #9. The probabilities for the measurement of which observable are stated there? Also the numbers don't add to 1 but to ##1/2(1-\cos \theta)+(|\alpha|^2+|\beta|^2)\cos \theta=1/2(1+\cos \theta)##.

1. I picked my example from here:
2. The ##- (\frac 1 2 )\cos \theta## is twice. Hence, it cancels ##+\cos \theta##

3. If the direction of measurement is at an angle ##\theta## w.r.t. the original direction vector the probability of spin up or down in that new direction (it can be alternatively considered as the basis rotation w.r.t the original basis if i understand correctly)..

PeroK
Homework Helper
Gold Member
2020 Award
The two-spin state in #1 is
$$|\Psi \rangle=\frac{1}{\sqrt{2}} (|1/2,1/2 \rangle + |-1/2,-1/2 \rangle).$$
This is a state belonging to total spin ##S=1##, i.e., the spin-triplet state and thus is not rotationally invariant.

I also do not understand the notation in #9. The probabilities for the measurement of which observable are stated there? Also the numbers don't add to 1 but to ##1/2(1-\cos \theta)+(|\alpha|^2+|\beta|^2)\cos \theta=1/2(1+\cos \theta)##.
The probabilities in post #9 fairly obviously add to ##1##.

PeroK
Homework Helper
Gold Member
2020 Award
I also do not understand the notation in #9.
##p(\uparrow_n | \psi)## is the probablity of getting spin up in direction ##\vec n##, at an angle of ##\theta## to the z-axis, given the state ##\psi## in the OP.

vanhees71
Gold Member
The probabilities in post #9 fairly obviously add to ##1##.
Ok, then I don't understand the notation.

vanhees71
Gold Member
##p(\uparrow_n | \psi)## is the probablity of getting spin up in direction ##\vec n##, at an angle of ##\theta## to the z-axis, given the state ##\psi## in the OP.
There are two particles. Do you mean the spin of one of the particles or the total spin? It's hard to discuss, if the expressions are not well defined.

Let's do a well-defined example. I suppose that ##|\pm 1/2 \rangle## are the usual eigenstates of ##\hat{\sigma}_z## with eigenvalues ##\pm 1/2## (##\hbar=1##).

The two-spin state is ##|\Psi \rangle \langle \Psi|## with
$$|\Psi \rangle=\alpha |1/2,1/2 \rangle + \beta |-1/2,-1/2 \rangle, \quad |\alpha|^2+|\beta|^2$$
Now let's measure a spin component in direction taking an angle ##\theta## with the 3-direction. To be concrete, consider ##\phi=0## in the usual spherical coordinates. Then
$$\hat{\sigma}_{\theta}=\sin \theta \hat{\sigma}_x + \cos \theta \hat{\sigma}_z.$$
The eigenvectors are
$$|\chi_+ \rangle=\cos(\theta/2) |1/2 \rangle + \sin(\theta/2) |-1/2 \rangle$$
and
$$|\chi_- \rangle = -\sin(\theta/2) |1/2 \rangle + \cos(theta/2)|-1/2 \rangle.$$
The probability to measure ##+1/2## for the 1st spin thus is
$$P_+=\sum_{j =\pm 1/2} |\langle \chi_+,j|\Psi \rangle|^2.$$
Now
$$\langle \chi_+,1/2|\Psi \rangle=\alpha \cos(\theta/2), \quad \langle \chi_+,-1/2|\Psi \rangle=\beta \sin(\theta/2)$$
and thus
$$P_+=|\alpha|^2 \cos^2(\theta/2)+|\beta|^2 \sin^2 (\theta/2).$$
For ##P_-## we need
$$\langle \chi_-,1/2|\Psi \rangle=-\alpha \sin(\theta/2), \quad \langle \chi_-,-1/2 \rangle=\beta \cos(\theta/2)$$
and thus
$$P_-=|\alpha^2| \sin^2(\theta/2)+|\beta|^2 \cos^2(\theta/2).$$
I hope, I've not made a mistake, but at least now the probabilities clearly add to 1 :-)).

Last edited:
PeroK
Homework Helper
Gold Member
2020 Award
There are two particles. Do you mean the spin of one of the particles or the total spin? It's hard to discuss, if the expressions are not well defined.
Yes, sorry, the spin of one particle.

vanhees71
There are two particles. Do you mean the spin of one of the particles or the total spin? It's hard to discuss, if the expressions are not well defined.

Let's do a well-defined example. I suppose that ##|\pm 1/2 \rangle## are the usual eigenstates of ##\hat{\sigma}_z## with eigenvalues ##\pm 1/2## (##\hbar=1##).

The two-spin state is ##|\Psi \rangle \langle \Psi|## with
$$|\Psi \rangle=\alpha |1/2,1/2 \rangle + \beta |-1/2,-1/2 \rangle, \quad |\alpha|^2+|\beta|^2$$
Now let's measure a spin component in direction taking an angle ##\theta## with the 3-direction. To be concrete, consider ##\phi=0## in the usual spherical coordinates. Then
$$\hat{\sigma}_{\theta}=\sin \theta \hat{\sigma}_x + \cos \theta \hat{\sigma}_z.$$
The eigenvectors are
$$|\chi_+ \rangle=\cos(\theta/2) |1/2 \rangle + \sin(\theta/2) |-1/2 \rangle$$
and
$$|\chi_- \rangle = -\sin(\theta/2) |1/2 \rangle + \cos(theta/2)|-1/2 \rangle.$$
The probability to measure ##+1/2## for the 1st spin thus is
$$P_+=\sum_{j =\pm 1/2} |\langle \chi_+,j|\Psi \rangle|^2.$$
Now
$$\langle \chi_+,1/2|\Psi \rangle=\alpha \cos(\theta/2), \quad \langle \chi_+,-1/2|\Psi \rangle=\beta \sin(\theta/2)$$
and thus
$$P_+=|\alpha|^2 \cos^2(\theta/2)+|\beta|^2 \sin^2 (\theta/2).$$
For ##P_-## we need
$$\langle \chi_-,1/2|\Psi \rangle=-\alpha \sin(\theta/2), \quad \langle \chi_-,-1/2 \rangle=\beta \cos(\theta/2)$$
and thus
$$P_-=|\alpha^2| \sin^2(\theta/2)+|\beta|^2 \cos^2(\theta/2).$$
I hope, I've not made a mistake, but at least now the probabilities clearly add to 1 :-)).

So, just to be clear/reconfirm:

if Alice and Bob have 1 particle each and Alice decides to measure its particle (in a direction that has an angle ##\theta## with the original direction ##z##) the probability of both Alice and Bob getting a ##\uparrow \uparrow## and getting a ##\downarrow \downarrow## in new direction is given by the eqns in #9 correctly. right?

PeroK
Homework Helper
Gold Member
2020 Award
So, just to be clear/reconfirm:

if Alice and Bob have 1 particle each and Alice decides to measure its particle (in a direction that has an angle ##\theta## with the original direction ##z##) the probability of both Alice and Bob getting a ##\uparrow \uparrow## and getting a ##\downarrow \downarrow## in new direction is given by the eqns in #9 correctly. right?
No. Post #9 was for either Alice's particle or Bob's particle.

The probability that the spins are opposite in this direction for the Bell state ##\frac 1 {\sqrt 2}(\uparrow \uparrow + \downarrow \downarrow)## is:
$$\frac 1 2 \sin^2 \theta (1 - \cos (2\phi))$$
This shows that the state is not as rotationally invariant as you may have believed. In fact, in the y-direction, with ##\theta = \phi = \frac \pi 2##, we have total anticorrelation:
$$\frac 1 {\sqrt 2}(\uparrow_z \uparrow_z + \downarrow_z \downarrow_z) = -\frac 1 {\sqrt 2}(\uparrow_y \downarrow_y + \downarrow_y \uparrow_y)$$

No. Post #9 was for either Alice's particle or Bob's particle.

The probability that the spins are opposite in this direction for the Bell state ##\frac 1 {\sqrt 2}(\uparrow \uparrow + \downarrow \downarrow)## is:
$$\frac 1 2 \sin^2 \theta (1 - \cos (2\phi))$$
This shows that the state is not as rotationally invariant as you may have believed. In fact, in the y-direction, with ##\theta = \phi = \frac \pi 2##, we have total anticorrelation:
$$\frac 1 {\sqrt 2}(\uparrow_z \uparrow_z + \downarrow_z \downarrow_z) = -\frac 1 {\sqrt 2}(\uparrow_y \downarrow_y + \downarrow_y \uparrow_y)$$

Now I am terribly lost to the point of giving up any hope of understanding even the basics unless I have a complete formal background in physics. The original Q was regarding arbitrary ##α, β## and I assumed there is only one angle ##θ## (as described in the video). I have no clue about the other details (my background is CS), namely ##\phi## etc. Thank you for the help though..

PeroK
Homework Helper
Gold Member
2020 Award
Now I am terribly lost to the point of giving up any hope of understanding even the basics unless I have a complete formal background in physics. The original Q was regarding arbitrary ##α, β## and I assumed there is only one angle ##θ## (as described in the video). I have no clue about the other details (my background is CS), namely ##\phi## etc. Thank you for the help though..
The video considers rotation in a plane (defined by a single angle ##\theta##). The state has rotational invariance in a plane. But, nature is three dimensional (hence rotations in 3D are defined by two angles ##\theta## and ##\phi##). That Bell state does not have 3D rotational invariance.

And, in answer to your question, you lose all rotational invariance unless you have ##|\alpha| = |\beta| = \frac 1 {\sqrt 2}##.

PeroK
Homework Helper
Gold Member
2020 Award
PS the Bell state ##\frac 1 {\sqrt 2}(\uparrow \downarrow - \downarrow \uparrow)##, which represents the spin-0 singlet state, has full 3D rotational invariance - which the other Bell states do not.

PS the Bell state ##\frac 1 {\sqrt 2}(\uparrow \downarrow - \downarrow \uparrow)##, which represents the spin-0 singlet state, has full 3D rotational invariance - which the other Bell states do not.

Got it. But if we limit ourselves to a plane (which I was assuming as its in the video example hence the confusion) we can set ##\phi##=0 in #21 then from it we can get the probability of opposite spin. But, it doesn't answer about the probability of ##\uparrow \uparrow## and ##\downarrow \downarrow## each given arbitrary α, β.

PS Sorry for testing your patience.