Sunday, April 28, 2019

HW 5. Due Monday May 6. With solutions.

HW5 soln link
https://drive.google.com/file/d/1asYW5eZ73eUyvEtGpD9KUsHxAWnfuaWc/view?usp=sharing

I would recommend starting on this soon and working on it some each day so you have time to ruminate and ask questions as needed. I think that is what a lot of people are doing and having really good success with that approach. I don't believe that you can do this all at one time and really learn a lot. Just my 2 cents. 

Additionally, I think it will really help your success if you ask questions when you are stuck. You can ask questions here or by email. Actually, sometimes posting your question here and asking me by email is good. Engagement and discussion are encouraged for this assignment and all assignments.

1.  a) Show that \(\psi_k = e^{ikx}\) is an energy eigenstate of the Schrodinger equation (TISE) for a free electron: \( - \frac {\hbar^2}{2m} \frac {\partial^2} {\partial x^2} \psi (x) = E \psi(x) \) What is its energy eigenvalue, \(E_k\)?
 b) Also, show that \(\psi_k = e^{ikx} e^{-iE_kt/\hbar}\) is solves the Schrodinger wave equation for a free electron: \( - \frac {\hbar^2}{2m} \frac {\partial^2} {\partial x^2} \psi (x) = i\hbar \frac {\partial \psi(x)}{\partial t} \)

2. Consider a state of the form,
 $$\Psi(x,t) =\frac{1}{\sqrt{2\pi}}\int_{-\infty}^{+\infty}\phi(k) \, e^{ikx} \,e^{-iE_kt/\hbar} dk$$
This is similar to non-stationary states we have looked at before, e.g., in the previous homework, however, instead of involving just two energy eigenstates, it involves an infinite number of energy eigenstates.  Explain the nature and meaning of each of the 3 terms in this integrand, and why the integral is over k.

3. Let's consider the specific case of: \(\phi(k)= A_a e^{-(k-k_o)^2a^2}\).
a) Determine \(A_a\) using a normalization constraint* on \(\phi(k)\) and show that, at t=0, integrating the equation above with this specific \(\phi(k)\) leads to:
\(\Psi(x,0) =\frac{1}{(a\sqrt{2 \pi})^{1/2}} e^{-x^2/(4a^2)} e^{ik_o x}\).
b) Describe this wave-function. Check its normalization.
c) Sketch a graph of the real part of this wave-function for a=1 nm and \(k_o = 6.28 \: nm^{-1}\).
d) Sketch a graph of the imaginary part of this wave-function for a=1 nm and \(k_o = 6.28 \: nm^{-1}\).
e) Sketch a graph of the probability density.
f) I would suggest a domain of about -4 to 4 nm for your plots.

4. Kinetic energy. This problem is triple credit because it is very important.
a) Calculate the expectation value of the kinetic energy at t=0 for an electron in this state \(\Psi(x,0) =\frac{1}{(a\sqrt{2 \pi})^{1/2}} e^{-x^2/(4a^2)} e^{ik_o x}\).
b) Discuss the character and nature of the different parts of the kinetic energy using terms like velocity, momentum, confinement and .... (You could back to this question after doing the next problem.)

5.  Time dependence. To understand the nature of the kinetic energy, it is helpful to examine time dependence of an electron in this state. Perhaps we can see how fast the electron is moving. Using the same integral form of \(\Psi(x,t)\) and the same \(\phi (k)\). 
a) Show that integrating over k leads to:
 \(\Psi(x,t) =\frac{1}{(a\sqrt{2 \pi} (1+it/t_c))^{1/2}} e^{-(x-2 k_o a^2 t/t_c)^2/(4a^2(1+t^2/t_c^2)}e^{ik_o x/(1+t^2/t_c^2)} e^{ix^2t/t_c/(4a^2(1+t^2/t_c^2))} e^{-i(k_o a)^2t/t_c/(1+t^2/t_c^2)}\)
where  \(t_c=(\frac {\hbar} {2ma^2})^{-1}\) is a characteristic time scale for this wave-packet state.
b) Describe this wave-function or, if you prefer, you can describe \(|{\Psi (x,t)}|^2\).
For for a=1 nm and \(k_o = 6.28 \: nm^{-1}\):
c) Plot the probability density for this electron at t=0, at \(t=t_c\)  and at \(t= 2 t_c\).  (3 different graphs).  What is \(t_c\) in femto-seconds?
d) With what speed is the electron wave-packet is traveling?
extra credit: e) Guess and then calculate the value of the expectation value of x as a function of time. What is the value of the expectation value of x at t= 100 femto-seconds?
f) How does the "size" of the electron seem to be evolving with time?

6. Consider an electron, one electron, in the wave packet state:  \(\Psi(x,t) =\frac{1}{\sqrt{2\pi}}\int_{-\infty}^{+\infty} \phi(k)\, e^{ikx} \,e^{-iE_kt/\hbar} dk\),
where \(\phi(k)= \frac{\sqrt{a}}{(\pi/2)^{1/4}}  e^{-k^2a^2}\).
a) Calculate \(\Psi(x,0)\). Is it normalized correctly?
b) Graph  \(\Psi(x,0)\). What is its value at x=0?
c) Calculate the expectation value of the kinetic energy for this electron at t=0.
d) How does this kinetic energy EVal compare with that of the electron from problems 4-7?
e) extra credit: how does the size of this electron evolve as a function of time?

----------------
Some terminology for scattering:  This is an informal, non-technical version of some terminology.
free state, free electron state, which is usually short for free-electron energy-eigenstate:
This refers to an energy eigenstate of the potential V(x) =0 which has the form \(e^{ikx}\)
unbound state is kind of like a free state, except that V(x) is not exactly zero everywhere.
bound state: A state the diminishes exponentially at large x and is normalizable, like the energy-eigenstates of the HO or finite square well.
Any energy eigenstate should be classifiable as either bound or unbound. Free state lies within the unbound category.
---------------
7. Scattering. Consider an electron and an attractive delta-function potential, \(V(x)=-\alpha \delta (x)\). One can use boundary conditions and the TISE to construct incoming and outgoing waves in the region \(x<0\), and an outgoing "transmitted" wave in the region \(x>0\). (These names only really make sense in the context of wave-packet constructions, however, it is a useful mathematical exercise to construct an unbound state with these 3 parts. It may help contextualize this formalism when we learn to understand localization and movement of a free electron.) This single-electron psuedo-state is an energy eigenstate, but is not normalizable so the coefficients in front of each of the three terms can't be called normalization parameters like we did for bound states. Let's call them pre-factors and set the prefactor of the incoming wave equal to 1. (This is an arbitrary convention, but I imagine that it will be useful.) In this nomenclature, the state of a single electron which extends through all space can be written as:\( e^{ikx} + B e^{-ikx} \:\: x<0\)
\( C e^{ikx} \:\: x>0\)
a) Find B and C as a function of the strength of the delta-function and k.
b) Calculate the reflection and transmission coefficients, R and T.
c) Graph R & T as a function of k. Over what range of k is T less than 0.5? Over what range of k is R greater than 0.5?
d) Over what range of k is T greater than 0.5? Over what range of k is R less than 0.5?
e) At what value of k is R=T=0.5?
f) For \(\alpha= 76\: meV*nm\), what value of k in \(nm^{-1}\) leads to T=0.5? What is 1/k? How does this particular value of 1/k compare with the length scale associated with the bound state of this attractive delta-function potential?
f) T approaches one at very high energy. Show that T can be viewed as crossing over from a low energy region in which T is less than 1/2, to a high energy region where T is greater than 1/2 . What would you pick as the crossover energy? How does that compare with the energy of the ground state of this potential?
g) Now let's consider a positive delta function. Do all the same things and see if they are different or the same. This might not take much extra work.
---------------

8. optional. If you are really into scattering, it might be interesting to do the same things as problem 10 for a positive or negative positive square well.I think there may be "resonances" where T=1 at particular k values. You'll have to define a notation with more terms...

44 comments:

  1. For #3, I'm getting a sqrt(pi) from the gaussian integral, which leaves pi^(1/4) in the numerator, not the denominator.

    ReplyDelete
    Replies
    1. This comment has been removed by the author.

      Delete
    2. Yes. I think the error is missing a factor of \(1/\sqrt{\pi}\) in front of \(\phi(k)\)

      Delete
    3. Thanks, I changed that. Does that look okay now?

      Delete
    4. I think part 8 and 9 need adjustment as well.

      Delete
  2. I think the best way to fix this is to make the coefficient of phi(k) sqrt(a)*(2/pi)^(1/4), then use the symmetric version of the Fourier transform which has the 1/sqrt(2pi) in front. This way both phi(k) and psi(x) are normalized. It is a little easier to calculate the expectation value of KE in momentum space, so it would be nice to have the normalized phi(k).

    ReplyDelete
    Replies
    1. Do you mean \(k\) space not momentum space (unless you have \(\hbar=1\) which is preferable for sure :) )? Anyway, thats a good point!

      Also, I suppose \(\phi(k)\) representation is useful only when involving various quantities involving \(\hat{p}\). The time evolution operator seems complicated in position representation (https://physics.stackexchange.com/questions/327507/convert-time-operator-from-momentum-space-to-position-space).

      Delete
    2. Oh right, yes I meant k space.

      Delete
    3. True, but you can also do that integral in real space and for problem 4-6 I think that is easier and also very revealing.

      Delete
    4. Your way is probably more general and powerful, and better in that way, but not needed for the t=0 calculation.

      Delete
  3. I'm a bit unclear on #3, part a, I tried normalizing the phi(k) function from -infinity to infinity to solve for Aa. What I got from the normalizing was { 1 = sqrt(pi/2)*(abs.val(a)/a^2)*(Aa)^2 }, so I'm not sure if I did the gaussian integral correctly?

    ReplyDelete
    Replies
    1. a is positive, so you can simplify |a|/a^2

      Delete
  4. For #7, is the h-bar in the numerator of tc supposed to be h-bar squared?

    ReplyDelete
    Replies
    1. You can check if it has dimensions of time.

      Delete
    2. I think it does have units of time as it is. You can multiply top and bottom by c^2 to turn the m into eV. Then write the left-over c as 3 x 10^17 nm/sec

      Delete
  5. For anyone interested, there is an alternate way of calculating the expectation of KE that is usually easier than the standard method. Using integration by parts, and the fact that psi is normalized (goes to zero at +/- infinity), there is a one line proof that the integral of (psi*)(-d^2/dx^2)(psi) is the same as the integral of |dpsi/dx|^2. This often lets you avoid using the product rule, or at least you only have to use it once rather than twice.

    ReplyDelete
    Replies
    1. Here that is again in mathjax:

      \(\int_{-\infty}^{\infty}\psi^*(-\frac{\partial^2}{\partial x^2})\psi dx=\int_{-\infty}^{\infty}|\frac{\partial \psi}{\partial x}|^2dx \)

      Delete
    2. Do you have a link to a proof for this?

      Delete
  6. I have a question regarding 3a. When the question says to apply the normalization condition, my instinct is to normalize \(\Psi\), not \(\phi(k)\). I see why \(<\Psi | \Psi> = 1 \), but why necessarily does \(<\phi(k) | \phi(k)> = 1 \). Thank you!

    ReplyDelete
    Replies
    1. Just like \( |\psi(x)|^2dx \) is the probability of finding the particle between \(x\) and \(x+dx\), \( |\phi(k)|^2dk \) is the probability of measuring the particle's k value between \(k\) and \(k+dk\), i.e. the probability of measuring its momentum between \( \hbar k \) and \(\hbar (k+dk) \). Since \( |\phi(k)|^2 \) is a probability density, it too must be normalized.

      This gives you another way to calculate \( < p^2> \). The momentum operator in \(k\)-space is \( \hbar k\), so \( < p^2>\ \)= \( \int_{-\infty}^{\infty} \phi*(\hbar k)^2 \phi dk = \hbar^2\int_{-\infty}^{\infty} k^2|\phi|^2 dk \).

      Delete
    2. Thank you, Lance, for the great explanation!

      Delete
  7. This comment has been removed by the author.

    ReplyDelete
  8. For #4 part a, does the kinetic energy of the time-independent electron in a particular state the normalization probability of the wave function and the 2nd order derivative w/ respect to x:
    = (-ℏ^2/2m)*(d^2/(dx^2))*(|ψ(x,0)|^2)

    ReplyDelete
    Replies
    1. Good question, but that is not the right way to calculate it. 1st thing, and this may be related, it is not a time-independent electron. We are just evaluating the EVal of the KE at a particular time for a time-dependent, moving electron. (So that is sort of like a ball moving through the air and asking what its KE is at a particular time.) Second, you don't want to create the probability density first and then take the derivative of it. That is a common misconception, but it is absolutely incorrect both conceptually and mathematically. If anyone has ever told you that the prob density tells you all about the electron, that is wrong. You can calculate the KE expectation value the same way we have in the past.

      Delete
    2. This comment has been removed by the author.

      Delete
    3. The method you that you are referring to would be to take the derivative first before multiplying by the conjugate of the function?

      = \( \-(hbar^2/(2m)* integrate from -infinity to infinity spi*(x)*(d^2/dx^2)*spi(x)dx \)

      does this concept still deviate from your previous suggestion?

      Delete
    4. The way we did it in HW2, problem 2.
      (2. a) Calculate the expectation value of the kinetic energy for an electron in the ground state of a 1D HO potential. )

      Delete
  9. Hi, I'm having trouble on the integral for problem 6 part a) especially for the phi(k) integral. The expression I inputed on wolfram reads: 'integrate from -infinity to infinity e^(-k^2*a^2)e^(ikx)dk' but I do not get a valid response from the integral. Did anyone approached the phi(k) integral otherwise?

    ReplyDelete
    Replies
    1. I don't think Wolfram alpha will be able to handle 3 undefined symbols, k, a and x. How are other people doing that? I think this may be related to an integral that you have covered in the 116 series? Can anyone confirm that?

      Delete
    2. If you did #3 already, notice the phi(k) for 6 is the same except \(k_0=0\)

      Delete
  10. This comment has been removed by the author.

    ReplyDelete
  11. I am pretty stuck on part 5a. Are we supposed to treat \(E_k\) as a constant? Are we supposed to integrate only the real part?

    Assuming the latter but not the former, I am stuck at
    \(\Psi (x,t) = \frac{\sqrt a}{(2\pi^3)^{1/4}} \int_{-\infty}^{\infty} e^{-(k-k_0)^2 a^2} cos(kx-\frac{\hbar k^2 t}{2m}) dk \)

    ReplyDelete
    Replies
    1. You need to integrate the whole thing. If you leave the \(e^{i(kx-E_kt/\hbar} \) as an exponential, you can combine the exponentials into one term that looks like \( e^{-Ak^2+Bk+C} \). You can relate this to a Gaussian integral.

      Note that A,B, and C can have i, x, or t in them, they are all constants as far as the integral is concerned.

      Delete
  12. What time is this homework due?

    ReplyDelete
  13. I am doing #6 and there is a question asking whether the wavefunction is normalized. I found that it was not, but the problem did not say if I should add the extra normalization constant or just continue without doing so. Any guidance? I checked my math a few times.

    ReplyDelete
    Replies
    1. What did you get for \(\Psi(x,0)\)?

      Delete
    2. I am thinking that you should get:
      \(\Psi(x,0) =\frac{1}{(a\sqrt{2 \pi})^{1/2}} e^{-x^2/(4a^2)}\).
      I am thinking that that is normalized.

      Delete
    3. My guidance is to add the extra normalization if needed, because you can't calculate a meaningful expectation value without having normalization.

      Delete
    4. I figured it out. I was using the gaussian integral value you gave several lectures ago, which was missing a root 2. It is normalized.

      Delete
    5. Good work! I have a request. Could you use a name to comment rather than unknown. It can be a made up name. Too many unknowns gets confusing sometimes. Your questions are helpful.

      Delete
  14. I am going to sign off now. If other people can answer any questions that arise, that will be much appreciated.

    ReplyDelete
  15. This comment has been removed by the author.

    ReplyDelete

Midterm 2 solutions

Here are solutions to midterm 2. I see that I did not write it in the solutions, but it is helpful for sp2 type integrals to recall that \(\...