# QM: Arbitrary operators and their eigenstates

1. Feb 2, 2009

### Niles

1. The problem statement, all variables and given/known data
Hi all.

When riding home today from school on my bike, I was thinking about some QM. The general solution to the Schrödinger equation for t=0 is given by:

$$\Psi(x,0)=\sum_n c_n\psi_n(x),$$

where $\psi_n(x)$ are the eigenfunctions of the Hamiltonian. We know that the probability of $\Psi(x)$ collapsing to one of the eigenstates is given by the absolute square in front of that particular eigenstate.

Now my question comes: Lets say we have an arbitrary operator Q: $Q\psi_n = q_n\psi_n$, where qn are the eigenvalues. Now what do I do if I want to find the probability of $\Psi(x)$ collapsing to one of the eigenstates of the operator Q upon measurement of the observable Q?

My own attempt is that we write $\Psi(x)$ as a new linear combination of the eigenstates of Q. But to me this seems very difficult.

I hope you can shed some light on this. Thanks in advance.

Best regards,
Niles.

2. Feb 2, 2009

### meopemuk

That's exactly what you should do.

3. Feb 2, 2009

### Niles

Great, that's really good.

I have a final question. Lets say we wish to find the expectation value of the operator Q, when our particle is in the state $\Psi$. The expectation value of Q is given by:

$$\left\langle Q \right\rangle = \left\langle \Psi \right|Q\left| \Psi \right\rangle.$$

But does this imply that $\Psi$ must be given in the same basis as the operator Q?

4. Feb 2, 2009

### nrqed

This calculation can be done in any basis one desires. The result is independent of the basis used. If the basis is not made of the eigenstates of Q then the representation of Q will not be diagonal, but that does not matter for the final result.

5. Feb 2, 2009

### Niles

But the only demand is that both $\Psi$ and Q are given in the same basis, right?

6. Feb 2, 2009

### nrqed

Yes, to carry out the explicit calculation, Q and Psi must be written using the same basis. That's right.

7. Feb 2, 2009

### Niles

Thanks for all your help, both of you.

8. Feb 3, 2009

### Niles

I have one final question, which is related to what I wrote in my first post:

Why is it that we choose to write it as a linear combination of the stationary states of the Hamiltonian, and not some other operator? Of course, as we talked about in this thread, we can express it as a linear combination of eigenfunctions of an arbitrary observable operator, but why have we chosen the Hamiltonian as the "default" operator?

9. Feb 3, 2009

### meopemuk

The choice of basis is a matter of convenience. For many problems (in atomic or nuclear physics) the use of Hamiltonian eigenfunctions makes calculations simpler than in other bases.

10. Feb 4, 2009

### Niles

Ahh, I see.

In my book, the expectation value of momentum is defined as:

$$\left\langle p \right\rangle = \int {\psi ^* (x)\left( {\frac{\hbar }{i}\frac{d}{{dx}}} \right)} \,\psi (x)\,dx,$$

where * denotes the complex conjugate. I have written the wavefunction as a function of x to emphasize that it is the wavefunction expressed in the x-basis (i.e. the basis of the position operator).

This must imply (according to our above discussion) that the momentum-operator in the x-basis is given as:

$$\hat p = {\frac{\hbar }{i}\frac{d}{{dx}}}.$$

Questions:

1) Am I correct that it really is the momentum operator given in the x-basis?

2) If yes, then how does one find this result?

I hope I am not annoying you with my questions. Thanks in advance.

Best regards,
Niles.

11. Feb 4, 2009

### nrqed

As meopemuk said, one can use any basis in principle. However, there is something special about using the eigenstates of the Hamiltonian. The hamiltonian is the generator of time translation. What this means is that it is the eigenstates of the Hamiltonian that have simpel time evolutions. They simply evolve as $\psi_n(x) exp(-iE_n t / \hbar)$. Eigenstates of other operators do not have simple time evolutions (unless they commute with the Hamiltonian and hence share common eigenstates). So if we want to know how a wavefunction evolves with time (let's say we are given it at t =0 and want to know what it wil be at a later time), we need to expand it over the eigenstates of the Hamiltonian in order to be able to write down the time dependence.

This is what makes the eigenstates of the Hamiltonian "special" .

12. Feb 4, 2009

### nrqed

I fixed the TeX

13. Feb 4, 2009

### xboy

Thanks a lot, nrged [:

14. Feb 4, 2009

### xboy

I should add that in the second equation, $$\hat K$$ is the generator.

15. Feb 4, 2009

### Niles

Great, thanks for taking the time to answer all my question to all three of you. It is very nice to get answers and derivations to these questions. I really appreciate it.